feat(android): tap to play, hold to select

Phase 1 of #63, and the design the issue asks for as one piece is
.planning/plans/active/019-android-touch-model.md.

**A finger has no second button and no modifier keys**, so the primary
action has to be the primary gesture: tap plays the row, and the hold
that opened a context menu now enters selection mode with that row
selected.

Three decisions in it, and two diverge from the report.

**The predicate is the pointer, not the platform or the viewport.**
`pointerType === 'touch'`, per event, which is already how long-press.ts
decided and is the only such test in the frontend. This is #64's rule --
named after the capability -- and it carries #64's warning: keyed on a
width, an Android *tablet* at 600px gets click-selects/double-click-plays
on a touchscreen, which is the inversion this issue exists to fix, on
the platform it exists for. A touchscreen laptop cannot be described by
a width at all. Per event, a mouse keeps desktop semantics on the very
same row, and there is no second declaration of what a phone does.

**There is no double-tap, and the number is why.** The report asks for
single tap to play *and* double tap for the menu. Those cannot both be
honoured: the first tap of a double tap is indistinguishable from a
single tap until the interval expires, so "tap plays" becomes "tap
waits". Measured on the device, the play command to TrackChanged is
155/123/85/56/91 ms -- median ~100 -- and the app's own
DOUBLE_CLICK_GRACE_MS is 250. That is 3.5x the primary interaction,
250ms of it spent deliberately doing nothing, on every track anyone
plays, to reach a menu the hold already reaches. So the menu and the
selection action bar are the same surface, which is also the platform's
convention and removes a concept rather than adding one.

**Tap-to-play and selection mode ship together**, because splitting
them is a regression dressed as an increment: a touch user selects by
tapping today and acts through the long-press menu, so moving tap to
play on its own would leave a window with no way to select forty tracks
at all.

**What lets this reassign the hold without touching one of the fourteen
context menus**: the layer announces `yj-tap` / `yj-long-press`
(composed, cancelable) and acts on nothing. A component claims one with
preventDefault. An **unclaimed long press still becomes a
`contextmenu`**, so the card grids, Explore, the playlist rows and
every other menu behave exactly as they did, and only lists that opt in
get selection mode. An unclaimed *tap* does nothing at all and the
click follows normally, which is what leaves every button in the app
alone -- only a claimed tap has its click swallowed, or playing a track
would also select it.

**And the device found the one thing no browser tier can see.**
Chrome 113's Android WebView fires its own `contextmenu` on a long
press. long-press.ts stood down when a trusted one arrived, which was
right while both paths ended in a context menu; they no longer do, so
standing down means the gesture silently does the *old* thing.
Measured, before the fix, holding a track row:

    {"log":["contextmenu isTrusted=true"],
     "state":{"bar":null,"menuActive":true,"selected":1}}

`yj-long-press` was never announced, the menu opened, and all 26 tests
passed -- dispatched pointer events do not make a browser synthesise
one. So the native event is a **trigger, not a competitor**: the
gesture is announced from it and only a claim suppresses it. Unclaimed
it propagates untouched, which is the same "browser wins" outcome
reached by asking instead of assuming.

The tier could not find that and can hold it, because this module has
always told its own events apart by identity rather than isTrusted, so
an untrusted one from a test takes exactly the browser's path.

Verified on the device by *performing* the gestures rather than
describing the page -- `adb shell input tap` and `input swipe x y x y
700` reach the WebView as real pointer events, which is new here and is
written down in the plan with the pixel mapping. Tap plays; a hold
raises the bar with one selected and no menu; a tap toggles to two,
back to one, and the mode ends with the last row; an album card still
opens its context menu.

29 new tests. The e2e spec is rewritten to assert **both** halves --
the row selects, and a card elsewhere still opens the real menu --
because a spec that only checked the row would pass on a build that had
silently broken the other thirteen.

Phases 2-4 (swipe to queue, the other three surfaces, and what #67
inherits) are in the plan and not in this commit.

.pi/skills/yellowjacket-dev/references/android-tier.md

@@ -8,13 +8,29 @@ This tier answers "does the phone build run", nothing else. It is not a
 spec tier, it does not run in CI, and the app is not a usable Android
 player yet (plan 015 says why, at length).
 
-## Three facts that make failure invisible
+## Two facts that make failure invisible
 
-**Go's stdout does not reach logcat.** An Android app's fd 1 and 2 go to
-`/dev/null`. Every `slog` line the app writes is discarded — including
-the one naming the error it is about to exit on. `setprop
-log.redirect-stdio true` does not help: it redirects the *Java*
-runtime's `System.out`, and the Go code is a c-shared native library.
+There were three. The first was that **Go's stdout does not reach
+logcat** — an Android app's fd 1 and 2 go to `/dev/null`, so every
+`slog` line the app wrote was discarded, including the one naming the
+error it was about to exit on. That is fixed (#160):
+`backend/androidlog` is a `slog.Handler` over `__android_log_write`,
+selected in `main()` by build tag, and the app's whole diagnostic
+stream now arrives under the `yellowjacket` tag, which `make
+android-logs` filters for.
+
+What remains true about it is the part that misleads: **`setprop
+log.redirect-stdio true` still does not help**, because it redirects
+the *Java* runtime's `System.out` and the Go code is a c-shared native
+library. Nothing that reaches logcat here does so through stdout, so
+anything printed with `fmt.Println` is still lost. Log with `slog`.
+
+The tag is a fixed string rather than the application id, and that is
+load-bearing rather than tidy: the debug build carries
+`applicationIdSuffix ".dev"` so it can be installed beside the release
+app, and it is the only build whose WebView can be inspected — so a tag
+derived from the id would be filtered out on the one build anybody
+debugging this app is running.
 
 **`os.Exit` is a silent death.** `main()` ends several failure paths in
 `os.Exit(1)`. From Android's side that is a process that vanished:
@@ -34,7 +50,10 @@ the wrong question. `make android-smoke` asks the right one — is it the
 The tell, once you know it: `I/WailsBridge: Wails bridge initialized`
 followed immediately by a new pid doing the same thing. That means the
 native library loaded, the JNI bridge came up, Go's `main()` ran, and
-`main()` left. Work backwards through its `os.Exit(1)` paths.
+`main()` left. Work backwards through its `os.Exit(1)` paths — and
+since #160, **read the `E/yellowjacket` line above it first**, because
+every one of those paths logs the error before it exits. That line is
+what #52 spent months without.
 
 ## What to run
 
@@ -524,6 +543,61 @@ Four things about it, each of which costs an hour if met cold:
   app.yellowjacket.dev android.permission.READ_MEDIA_AUDIO` (and
   `POST_NOTIFICATIONS`) ahead of the launch skips it.
 
+### Getting the app into a state worth measuring
+
+A fresh install is **not** a neutral starting point, and three things
+about it will each cost you a measurement.
+
+**It downloads the real catalog.** `YJ_CORE_INDEX_URL` is stubbed in
+`dev-headless.sh` and in CI and is *real* here, so the app spends its
+first minutes fetching ~0.6 GB and `job-band` is **103px of a 439px
+screen** while it does. Every vertical number taken in that state is
+wrong -- one #51 measurement had the album art at 0px and it was
+entirely this.
+
+`__yj.call("explore.Service.StopIndexBuild", [])` stops it and returns
+cleanly. **It then starts again within seconds.** So stop it
+*immediately before* the measurement rather than once at the beginning,
+and check `jobs.Service.GetJobs` afterwards -- an empty array is the
+only proof. `jobs.Service.ClearFinishedJobs` tidies the finished rows
+that otherwise keep the band open.
+
+**A library added over the bridge does not dismiss the first-run
+wizard.** `library.Library.AddLibrary` works and scans, but the wizard
+checks for an existing library once, on mount, and its "Get Started"
+button gates on a directory chosen *in the wizard* -- so it stays up
+with a correctly disabled button over everything you are trying to
+measure. Nothing is broken; reload the page and it is gone. This reads
+exactly like a tap being swallowed, which is the expensive part.
+
+**Scoped storage decides where the music can be.** `/sdcard/Music/...`
+plus `pm grant <pkg> android.permission.READ_MEDIA_AUDIO` works and
+`AddLibrary` takes the plain path; a push into
+`/sdcard/Android/data/<pkg>/files/` looks like it worked and then is not
+there. Some builds additionally want
+`appops set <pkg> MANAGE_EXTERNAL_STORAGE allow`, and until they have it
+the app opens the *system* "All files access" screen on launch -- so
+`dumpsys window | grep mCurrentFocus` naming `com.android.settings` is
+that, not a crash.
+
+### A note on quoting `make android-eval`
+
+`EXPR='...'` is a single-quoted shell word, so anything with a quote or
+an apostrophe in it -- a file path like `Blazo, 49'ers - ...`, or a
+snippet containing a string literal -- breaks in a way that reads as a
+JavaScript error. Put the expression in a file and pass it positionally:
+
+```bash
+node ./scripts/android-eval.mjs "$(cat /tmp/probe.js)"
+```
+
+That is the same script `make android-eval` wraps, so nothing is lost.
+Two things worth knowing about it: it does **not** await a promise, so
+an async call has to park its result (`window.__r = ...`) and be read
+back in a second eval; and the shim from the section below is lost on
+every reload and every app restart, along with the devtools socket,
+whose name carries the pid.
+
 ### Calling a binding on the device
 
 **The runtime call does not go over HTTP on Android**, and this is worth

.planning/NOTES.md

@@ -4785,3 +4785,130 @@ broken build. The specs assert the *mechanism* — that the surface is a
 native `<dialog>` at phone width — which is the same move
 `queue-as-a-screen.spec.ts` makes about containment and for the same
 reason.
+
+## Now Playing was not drawing a small square, it was drawing a crop (measured 2026-08-21, TLP301 / Chrome 113 / 424x439)
+
+#172 handed #51 a design question — whether the album art gets a floor
+with the block scrolling, or whether the screen reflows below some
+height. Measuring it first turned up a defect underneath the question,
+and the defect is bigger than the phone.
+
+**The art was never square.** `aspect-ratio` is specified not to
+re-derive the width when `max-height` clamps the height — unlike an
+intrinsic ratio, which CSS2.1 10.4 preserves under both bounds. So
+`width: min(100%, 60vh)` made the width definite, the ratio derived a
+height from it, `max-height: 100%` clipped that height, and the width
+stayed where it was. `object-fit: cover` then cropped a square cover
+into the band. On the device: **264x53**, a 5:1 strip. #172's own table
+called it "39px of art" and the missing half is that those 39px were
+263 wide.
+
+**And it is not only the phone.** The leftover exceeds the width only
+above ~843px of viewport, so every height from ~500 to ~843 drew a
+crop too — most phones, and any short window. The e2e spec written for
+this fails on the old build at 424x439 (263x39), 390x700 (358x315) and
+900x500 (300x36), and *passes* at 412x869, which is the boundary
+falling exactly where the arithmetic says it should.
+
+**Both maxes with auto sizes is the whole fix**, and it was chosen by
+asking Chrome 113 rather than by reasoning: a probe shadow root at
+column heights of 288, 300, 451, 600 and 800 measured four candidate
+rules. `max-width/max-height: 100%` with `width/height: auto` is square
+at all five; the shipped rule cropped at four; `aspect-ratio` on the
+box cropped at the tallest. A corollary that makes it free: `auto` will
+not upscale past the natural size, and the largest tier `saveCoverArt`
+keeps is 400px, so nothing is lost by never exceeding it.
+
+**The placeholder cannot use that rule and needed its own**, which is
+the part that would have shipped broken. It is not a replaced element,
+so with no intrinsic size auto/auto collapses it to its icon —
+measured at **13x58**, neither square nor the art's size. Three things
+about the rule it did get:
+
+- It is driven from the **height**, which is the axis that binds
+  everywhere this view is reached from.
+- A flex item's automatic minimum is its content, so without
+  `min-width: 0` the icon's own width becomes a floor and the box goes
+  wider than it is tall the moment the row is shorter than the icon —
+  which is exactly the state a job band puts this screen in.
+- **A non-replaced box cannot express "the largest square that fits" at
+  all**, because whichever max clamps does not re-derive the other. The
+  height-driven rule alone went **380x484** at 412x869 — a tall phone,
+  #51's other named device — and `max-height: calc(100vw - 2rem)` is
+  what closes it. That is sound here for the reason `60vh` was not: this
+  is a phone-width detail view, so its content box really is the
+  viewport less the host's own gutters, and it is a *max*, so if that
+  ever stopped being true the failure is a square bounded early rather
+  than a crop. `rem` and not `em` — the box sets `font-size: 3rem` for
+  the icon, so `2em` there is 96px.
+
+**Then the design question, and the reflow is the answer.** The
+stacked layout's budget is fixed — 48px of header, 143px of transport
+since #64, 78px of names, 68px of padding and gaps — so the art gets
+`height - 386`, which is 53px at 439. A floor on the art scrolls the
+transport off the bottom, and "controls never scroll off" is #51's own
+Direction and plan 018's promise. So below 500px the art and the names
+share a row, where the art is bounded by the row's height rather than
+by the column's leftover: **53px to 143px on the device**, measured on
+the shipped build, with nothing scrolling and the transport untouched.
+
+**500 is where the two layouts cross, not a round number.** In a row
+the art is `height - 296` and the names get what is left of 392px, so
+the names hold 176px at exactly 500 and less above it; stacked, the art
+is `height - 386`, which passes 176px at 562. It is keyed on height
+alone rather than on the phone's width because it is an answer to
+vertical room — a 900x450 window has the same problem and the same fix.
+
+Two things the audit found that are *not* this, and are filed:
+**#186**, every control that is not the transport is under the 44px
+floor (the sort direction arrow is 28x21, and `search-trigger` — which
+exists only on a phone — is 40x40), and **#187**, the seek bar's drag
+target is 6px tall.
+
+**What the audit did not find is a reachability failure**, which is
+worth recording because it is the promise plan 018 makes. At 424x439,
+on every view -- the ten primary ones, the queue, `album-details`,
+`artist-details`, Downloads and Autotag -- `documentElement.scrollWidth`
+is 424 against a 424 viewport, no control sits outside a scrollable
+ancestor, and a hit test at each control's centre reaches the control.
+The width work of #57, #62, #55 and #59 holds; what was left was
+vertical, and it was this screen.
+
+**A number measured on the device is not a number CI can assert.** The
+spec's floor on the art's height passed here at 114 and failed in CI at
+**64**, and both are honest: this app is long-lived, so a job staged by
+an earlier spec is still on screen, and `volume-control` renders in a
+browser where it does not on Android. Both are chrome above and below
+the view and both move the leftover. That is the same trap the entry
+above about staged jobs describes, arriving as a *measurement* rather
+than as a stuck job. The assertion is the mechanism now -- in a row the
+art fills the row's height rather than being the leftover -- and the
+53-to-143 stays on the issue, where it was measured.
+
+The probe is worth keeping in mind for the next audit, because two of
+its three checks needed a second pass to mean anything. "Painted
+outside the viewport" flags a horizontally scrolling carousel -- the
+home shelves -- so the real question is whether a *scrollable ancestor*
+can bring the element back. And a hit test at a control's centre flags
+everything below the fold in a scroll container, so it only says
+something once the control is on screen. Both first drafts produced
+long lists of nothing.
+
+## Two traps that cost time on the device, both already written down (2026-08-21)
+
+Recorded because both are in `android-tier.md` and I met them anyway.
+
+**A fresh install downloads the real catalog**, so `job-band` is 103px
+of a 439px screen and every vertical measurement is wrong. Worse, it
+**restarts**: `explore.Service.StopIndexBuild` returns cleanly and the
+job is `running` again within seconds, so it has to be stopped again
+immediately before a measurement rather than once at the start.
+`YJ_CORE_INDEX_URL` is stubbed in `dev-headless.sh` and in CI and is
+real on a device.
+
+**The first-run wizard does not re-check for a library it did not
+create.** Adding one through `library.Library.AddLibrary` over the
+bridge leaves the wizard up with its "Get Started" button correctly
+disabled — it gates on a directory chosen *in the wizard*, and the
+existing-library check runs once, on mount. A reload clears it. Nothing
+is broken; it cost twenty minutes of believing a tap had been swallowed.

.planning/plans/active/019-android-touch-model.md

@@ -0,0 +1,253 @@
+# 019 — The Android touch model
+
+**Issue:** #63 (`Area/Library-UI`, `Kind/Feature`, `Priority/High`)
+**Depends on:** #60 (bottom-sheet menus) — closed, merged as PR #176
+**Relates:** #67 (inline links into the menu), #71 ("More" nav), #54
+(native feel), #5/#8 (selection, drag to queue — the desktop semantics
+being diverged from)
+**Status:** in flight.
+
+#73 puts #60 first in Phase 4 because it is "the presentation every
+other item needs", and this is the next one. The Direction on #63 asks
+for the interaction model to be designed as one piece before any of it
+is built, because it *reassigns an existing gesture* rather than adding
+one — `utils/long-press.ts` currently owns the 500ms hold, and every
+context menu in the app is downstream of it.
+
+This document is that design. Everything below is a measurement, or an
+argument for one of the choices #63 leaves open.
+
+---
+
+## The mapping
+
+| gesture | pointer is a finger | pointer is a mouse |
+|---|---|---|
+| single tap / click | **play the row** | select the row |
+| double | — | play the row |
+| long press (500ms) | **enter selection mode** | — |
+| right-click | — | context menu |
+| swipe right | **add to queue** | — |
+| drag | reorder / drag to playlist | reorder / drag to playlist |
+
+Three of those are #63's report unchanged. Two are decisions it left
+open, and one is a deliberate divergence.
+
+---
+
+## Decision 1 — the predicate is the pointer, not the platform
+
+#63 says "the row component needs a platform-aware interaction layer
+rather than shared handlers". It needs an interaction layer; it should
+not be platform-aware.
+
+**The question a row has to answer is not "am I on Android" or "is the
+viewport under 600px" but "what made this event".** `pointerType ===
+'touch'`, read off the event that is being handled, which is already
+how `long-press.ts` decides (`if (e.pointerType !== 'touch') return`)
+and is the only such test in the frontend today.
+
+This is #64's rule — the predicate is named after the capability, not
+the platform — and it carries #64's warning with it. Keyed on a width:
+
+- an Android **tablet** at 600px or more gets click-selects /
+  double-click-plays on a touchscreen, which is the exact inversion
+  this issue exists to fix, on the platform it exists for;
+- a **touchscreen laptop** cannot be described at all, because both
+  pointers are live in the same session on the same row;
+- and a narrow desktop window gets phone semantics with a mouse.
+
+Per event, all three are right for free, and there is no second
+declaration of what a phone does — the thing CLAUDE.md declines to add
+every time it comes up.
+
+**Measured, so this is not an assumption about the WebView.** On the
+reference device (TLP301, Android 14, WebView Chrome 113, 424x439),
+driving a real tap with `adb shell input tap`:
+
+```
+[["down","touch",78,94],["touchstart","touchstart",0,0],["up","touch",78,94]]
+```
+
+`PointerEvent` exists, `pointerType` is `"touch"`, `maxTouchPoints` is
+5, and `(pointer: coarse)` / `(hover: none)` both match.
+
+---
+
+## Decision 2 — there is no double-tap, and the number is why
+
+#63 asks for *single tap → play* **and** *double tap → context menu*.
+Those two cannot both be honoured. The first tap of a double tap is
+indistinguishable from a single tap until the interval expires, so
+"tap plays" necessarily becomes "tap waits to find out whether you
+meant something else, then plays". The app already owns that constant:
+`utils/explore-link.ts` holds a navigation for `DOUBLE_CLICK_GRACE_MS
+= 250` for precisely this reason.
+
+**What it would be added to, measured on the device.** Six runs, from
+the play command to the backend's `TrackChanged`:
+
+```
+155, 123, 85, 56, 91 ms          median ~100
+```
+
+So the app's primary interaction is ~100ms, and a double-tap
+discriminator makes it ~350 — **3.5x, of which 250ms is spent
+deliberately doing nothing** — paid on every track anyone ever plays,
+in order to reach a menu.
+
+It is also against the platform's convention, which counts for more
+than usual here because this is the phone build and nothing else:
+long-press is *how you select* on Android (Gmail, Files, Photos),
+double-tap is zoom or nothing, and a list's menu is either the
+long-press sheet or a per-row overflow.
+
+**So the menu and the selection action bar become the same surface**,
+which is the convention and removes a concept rather than adding one.
+Long-press selects the row it was made on and raises the action bar;
+the bar's actions *are* the context menu's actions, contextualised to
+whatever is selected — one row or forty. #60's bottom sheet stays
+behind it as the overflow, so `contextMenuStyles`, `MenuKeyboard` and
+`menu-surface` are reused rather than reimplemented.
+
+---
+
+## Decision 3 — tap-to-play and selection mode ship together
+
+The obvious phase order is "tap plays first, it is the smallest
+change". It is wrong, and the reason is a capability that exists today
+and is easy to miss.
+
+**A touch user can already multi-select**: tap selects (the desktop
+semantics, which a finger currently gets), and the long-press menu then
+acts on the selection. Move tap to play without shipping selection mode
+in the same change and there is a window — a release, if it lands — in
+which selecting forty tracks to add to a playlist is impossible on a
+phone. That is a regression dressed as an increment.
+
+So phase 1 is both, or neither.
+
+---
+
+## What the code looks like now
+
+| surface | how it binds | selection |
+|---|---|---|
+| `track-list` | delegated on the virtualizer: `click`, `dblclick`, `contextmenu`, `dragstart` | `SelectionController` |
+| `queue-panel` | delegated, same shape | `SelectionController` |
+| `playlist-details` | per row | `SelectionController` |
+| `smart-playlist-details` | per row | `SelectionController` |
+
+All four already share `SelectionController`, and all four resolve a
+row from an event by `data-index` / `data-file-path` on the row. So the
+gesture layer has one shape to talk to, and "selection mode" is a flag
+on the controller they already have rather than a fifth concept.
+
+`utils/long-press.ts` is one document-capture listener that synthesises
+a `contextmenu` — the seam that needed no component to opt in. **This
+plan keeps that shape and changes what the gesture means**, which is
+why it is a rewrite of that file rather than a second listener set: two
+document listeners both claiming the 500ms hold is the fault the file's
+own header warns about.
+
+---
+
+## Two measurements that decide the implementation
+
+**`touch-action` is `auto` on both the virtualizer and the rows.** With
+`auto` the browser owns panning on both axes, so a horizontal drag can
+be claimed as a scroll and our gesture ends in `pointercancel`
+mid-swipe. A row that wants a horizontal swipe has to declare
+`touch-action: pan-y`: the browser keeps the vertical pan (which is the
+virtualizer's scroll, and must stay native or the list stutters) and
+hands us the horizontal axis. This is the single most likely way for
+swipe-to-queue to "work in Chromium and not on the phone".
+
+**The row is 424x52 on the device**, so a swipe threshold in px is a
+fraction of a row height, not of a screen.
+
+**And the third one was found by building phase 1 and then running it**
+— it is not something any browser tier can report. Chrome 113's Android
+WebView **fires its own `contextmenu` on a long press**. `long-press.ts`
+stood down when a trusted one arrived, which was right while both paths
+ended in the same place; once a hold can mean selection mode they end
+in different places, and standing down means the gesture silently does
+the *old* thing. Measured, before the fix:
+
+```
+{"log":["contextmenu isTrusted=true"],
+ "state":{"bar":null,"menuActive":true,"selected":1}}
+```
+
+`yj-long-press` was never announced at all, the context menu opened,
+and all 26 tests in the component tier passed — dispatched pointer
+events do not make a browser synthesise a `contextmenu`.
+
+So the browser's event is a **trigger, not a competitor**: the gesture
+is announced from it, and only a component that claims it suppresses
+the native menu. Unclaimed, it propagates untouched. That is the same
+"browser wins" outcome, reached by asking instead of assuming — and
+verified both ways on the device, a track row entering selection mode
+and an album card still opening its menu.
+
+The tier could not *find* it and can *hold* it: a test cannot dispatch
+a trusted event, but this module has always told its own apart by
+identity rather than `isTrusted`, so an untrusted one from a test takes
+exactly the browser's path.
+
+---
+
+## A tier note: this one can be driven, not only measured
+
+`adb shell input tap|swipe` reaches the WebView as real pointer events,
+which the log above is evidence of. So for the first time the Android
+tier can *perform* the thing under test rather than describe the page
+afterwards — a long press is `input swipe X Y X Y 600`, a swipe right
+is `input swipe X Y X+N Y 120`.
+
+Device CSS pixels from device pixels, on this phone:
+`css = (device - 59) / 2.564` vertically, `css = device / 2.564`
+horizontally (measured from the tap above: 200,300 arrived as 78,94).
+
+This does not make the device a spec tier — it does not run in CI and
+`make ui-test` still has to carry the assertions. It makes "does the
+gesture actually fire on Chrome 113" answerable in seconds.
+
+---
+
+## Phases
+
+**Phase 1 — the seam, tap-to-play, selection mode.** `utils/
+touch-gestures.ts` replacing `long-press.ts`: pointer-typed
+recognition of tap / long-press / horizontal swipe, dispatched as
+composed custom events so a delegated listener in any shadow root
+still works. `SelectionController` gains a mode. `track-list` acts on
+tap and enters the mode on long press. The action bar.
+
+**Phase 2 — swipe right to queue**, with the `touch-action: pan-y`
+finding above and a reveal-and-snap affordance.
+
+**Phase 3 — the other three surfaces**, which is mostly wiring, since
+they already share the controller.
+
+**Phase 4 — what this leaves behind.** The inline `explore-link`s in a
+row are a single-click target inside a row whose single tap now plays;
+that conflict is #67's, and this plan should not pre-empt its answer
+beyond making tap-to-play win on touch.
+
+---
+
+## Open questions
+
+1. **Does selection mode have an escape other than the bar's own
+   close?** Back is the platform's answer and the shell already owns
+   the history stack (#6/#55). Pushing an entry for a *mode* rather
+   than a place is the same argument #55 settled for the overlaid
+   queue, and it should probably be settled the same way — but the
+   queue is a screen and a selection mode is not, so it wants its own
+   paragraph rather than an assumption.
+2. **Does a tap on a row's favourite icon still toggle it in normal
+   mode?** It is inside the row and the row now plays. It has to keep
+   working — it is a 44px target since #56 — so the gesture layer needs
+   the same "a control inside the row wins" rule the keyboard service
+   has for a focused control that owns a key.

CLAUDE.md

@@ -1462,6 +1462,42 @@ vary) wins, ours being told from theirs by **identity** rather than
 that ends the gesture is swallowed, keyed on the gesture rather than on
 a time window so the first tap on the menu it opened is not eaten too.
 
+**A control revealed by `:hover` is gated on the device having hover,
+and which way round depends on whether it is the only route to its
+action.** The gate itself is not optional: a touch long-press
+synthesises a hover state in the WebView, so every one of these flashed
+into view during the 500 ms hold above — a control appearing because
+the user was reaching for a different one. Where the action is reachable
+another way the control is **absent** on a touch device (the home card's
+play button, #68; the queue row's remove, which the row's bottom-sheet
+menu carries since #60), and that is `display: none` outside
+`(hover: hover) and (pointer: fine)` rather than `opacity: 0` or
+`visibility: hidden`, both of which leave a button holding its hit area
+and its place in the accessibility tree. Where the control is the
+**only** route it is instead always visible under
+`@media not all and (hover: hover)` — `track-details`'s cover-art
+overlay and remove, `shortcut-capture`'s reset (#137) — because hiding
+it takes the action away entirely.
+
+**Always-visible is not the same as always-in-the-way.** The cover-art
+overlay is `inset: 0` at 50% black, which is fine as a hover state and
+is not fine as the permanent appearance of the artwork being edited —
+and it is only a *hint*, since `.cover-art-edit` carries the click and
+tapping the art always worked. Off hover it becomes a corner chip in
+the remove button's own language. The × beside it stays full-size,
+because that one really is the only route to its action.
+
+One thing to know before checking either: **no *committed* tier renders
+as a touch device.** CDP's `Emulation.setEmulatedMedia` does not reach
+the component tier's iframe, and the e2e projects are Desktop Chrome
+and Desktop Safari, neither of which has touch — a Playwright project
+using a mobile descriptor would report `hover: none`, so this is a
+choice not to carry one rather than a thing that cannot be done. So
+`hover-affordance.test.ts` asserts the *parsed stylesheet* — which rule
+sits inside which media query — and says so; the regression it exists
+for is someone hoisting a rule out of its query as a tidy-up, which
+nothing on a desktop renders differently.
+
 Three lists had no focused row to open a menu *from* — the queue panel
 and both playlist detail views — and gained a roving tab stop through
 `utils/roving-rows.ts`. **`track-list` deliberately does not use it**:
@@ -2033,9 +2069,26 @@ one that closes the queue — the reported defect moved one press later,
 which looks exactly like a press that did nothing.
 
 **And the way out is 44px on a phone.** With the panel spanning the
-whole width the scrim has no uncovered pixels at all, so the close
-button is the only pointer route out of a full-screen surface; it was
-**25×21px**.
+whole width there is no scrim there at all, so the close button is the
+only pointer route out of a full-screen surface; it was **25×21px**.
+
+**The scrim is drawn only where it can be tapped** (#171). Below 600px
+`.panel-content` is `width: 100%`, so the scrim sat entirely underneath
+an opaque panel — measured at 424×439, host, panel and scrim all
+424×318 — dimming nothing and dismissing nothing while wearing
+`cursor: pointer`. #24's tap-outside-to-close cannot exist on a surface
+with no outside, and the screen above is what answers it instead: back,
+and a 44px close button. The alternative — a gutter, which is the
+drawer pattern — was declined, because it buys the affordance by taking
+width off a full-screen surface on a 424px viewport. Two things about
+it are load-bearing. Its **existence** is `matchMedia`, not
+`display: none`, on `job-band`'s rule: a hidden scrim is still an
+element carrying the dismissal handler. And **the 600–899 band is
+untouched**, where the panel is a 320px column of a wider content area
+and the scrim has real uncovered pixels — which is why the e2e half
+asserts *absence* at 424×439 rather than clicking, since a phone-width
+case that clicks the scrim's centre hits the panel and passes on the
+broken build.
 
 What this does **not** fix is `page-header` overflowing on its own:
 at 900×600 "New Smart Playlist" is still clipped to 114 of 162px with
@@ -2067,6 +2120,54 @@ toggled from `index.ts` would be a second expression of the same fact.
 The view therefore carries its own queue button, because that button
 lives in the bar it hides.
 
+**And below 500px of height its art and its names share a row** (#51).
+The stacked arrangement's budget is fixed — 48px of header, 143px of
+transport since #64, 78px of names, 68px of padding and gaps — so the
+art gets `height - 386`, which at the reference device's 424x439 is
+**53px**: the one thing a Now Playing screen exists to show, smallest
+on it. #172 named the two ways out and this is the second, because the
+first — a floor on the art with the block scrolling — scrolls the
+transport off the bottom, and *controls never scroll off* is #51's own
+Direction and plan 018's promise. Sideways the art is bounded by the
+row's height instead of by the column's leftover: **53px to 143px**,
+measured on the device, nothing scrolling, the transport untouched.
+
+Three things about it are load-bearing.
+
+**500 is where the two layouts cross rather than a round number.** In a
+row the art is `height - 296` and the names get what is left of 392px,
+so the names hold 176px at exactly 500 and less above it; stacked, the
+art is `height - 386`, which passes 176px at 562. Below 500 the row is
+the bigger art *and* the readable one — above it the column is, which
+is why a tall phone keeps the arrangement it has. It is keyed on height
+alone and not on the phone's width, because it answers vertical room: a
+900x450 window has the same problem and the same fix.
+
+**The art was not a small square, it was a crop, and that was never
+only the phone.** `aspect-ratio` is specified not to re-derive the
+width when `max-height` clamps the height — unlike an intrinsic ratio,
+which is preserved under both bounds — so a definite `width: min(100%,
+60vh)` kept its width while the height was clipped, and `object-fit:
+cover` cropped a square cover into the band: **264x53** on the device.
+The leftover exceeds the width only above ~843px of viewport, so every
+height from ~500 to ~843 drew one too. `max-width`/`max-height: 100%`
+with `width`/`height: auto` is the fix and is the replaced-element
+path; it also never upscales past the natural size, and the largest
+tier `saveCoverArt` keeps is 400px, so nothing is lost.
+
+**The placeholder needs its own rule, and a non-replaced box cannot
+express this one.** With no intrinsic size, auto/auto collapses it to
+its icon (13x58, measured). It is driven from the height instead, with
+`min-width: 0` because a flex item's automatic minimum is its content —
+without it the icon's width becomes a floor the moment the row is
+shorter than the icon, which is exactly what a job band does to this
+screen. And since whichever max clamps does not re-derive the other, a
+height-driven box goes **380x484** on a tall phone; `max-height:
+calc(100vw - 2rem)` closes it, which is sound here for the reason
+`60vh` was not — this is a phone-width detail view, so its content box
+really is the viewport less the host's gutters, and it is a *max*, so
+the failure mode is a square bounded early rather than a crop.
+
 **The playing row is a shape, not a hue.** `track-list` and
 `queue-panel` draw a `::before` triangle in each row's own left
 padding, plus `aria-current` — before, both rows were a background tint

backend/androidlog/android.go

@@ -0,0 +1,55 @@
+//go:build android
+
+// The write itself, and nothing else. Everything decidable off a phone
+// is in androidlog.go; see the package comment for why.
+
+package androidlog
+
+/*
+#cgo LDFLAGS: -llog
+#include <stdlib.h>
+#include <android/log.h>
+*/
+import "C"
+
+import (
+	"log/slog"
+	"unsafe"
+)
+
+// The priorities in androidlog.go are android/log.h's own values, and
+// these are what says so. A constant expression that would be negative
+// does not compile as a uint, so a renumbered header fails the build
+// here rather than logging everything at the wrong severity -- which is
+// the failure that would otherwise be invisible, since logcat would
+// happily print whatever number it was handed.
+const (
+	_ = uint(C.ANDROID_LOG_VERBOSE - PrioVerbose)
+	_ = uint(PrioVerbose - C.ANDROID_LOG_VERBOSE)
+	_ = uint(C.ANDROID_LOG_DEBUG - PrioDebug)
+	_ = uint(PrioDebug - C.ANDROID_LOG_DEBUG)
+	_ = uint(C.ANDROID_LOG_INFO - PrioInfo)
+	_ = uint(PrioInfo - C.ANDROID_LOG_INFO)
+	_ = uint(C.ANDROID_LOG_WARN - PrioWarn)
+	_ = uint(PrioWarn - C.ANDROID_LOG_WARN)
+	_ = uint(C.ANDROID_LOG_ERROR - PrioError)
+	_ = uint(PrioError - C.ANDROID_LOG_ERROR)
+	_ = uint(C.ANDROID_LOG_FATAL - PrioFatal)
+	_ = uint(PrioFatal - C.ANDROID_LOG_FATAL)
+)
+
+// New returns the handler main() installs on Android.
+func New(opts *slog.HandlerOptions) slog.Handler {
+	return NewHandler(opts, write)
+}
+
+// write hands one line to liblog.
+func write(prio int, tag, msg string) {
+	cTag := C.CString(tag)
+	defer C.free(unsafe.Pointer(cTag))
+
+	cMsg := C.CString(msg)
+	defer C.free(unsafe.Pointer(cMsg))
+
+	C.__android_log_write(C.int(prio), cTag, cMsg)
+}

backend/androidlog/androidlog.go

@@ -0,0 +1,250 @@
+// Package androidlog routes slog to logcat.
+//
+// **An Android app's fd 1 and 2 go to /dev/null**, so every line this
+// app writes with slog is discarded on that platform -- including the
+// one naming the error it is about to os.Exit on. #52 is what that
+// cost: a process that vanished with no tombstone, no AndroidRuntime
+// stack and nothing in `logcat -b crash`, at Priority/Critical for
+// months, whose entire diagnosis was one slog.Error main.go was
+// already writing.
+//
+// The platform's own sink is __android_log_write, which is a handful
+// of cgo -- and cgo compiled by nothing `make lint` or `make test`
+// runs, since the only toolchain that builds the android tag is a
+// cross-compiler and the only thing that runs it is a phone. So the
+// split here is the one backend/mediacontrols/androidpayload.go makes,
+// pushed as far as it will go: **everything except the write itself is
+// in this file, untagged**. The priority mapping, the formatting, the
+// chunking and the handler's own attr and group bookkeeping are
+// ordinary Go that `go test` exercises on any platform; android.go is
+// fifteen lines that hand a string to liblog.
+package androidlog
+
+import (
+	"bytes"
+	"context"
+	"log/slog"
+	"strconv"
+	"strings"
+	"sync"
+)
+
+// Tag is what logcat labels these lines with.
+//
+// It is a constant of ours rather than the application id, because the
+// debug build carries `applicationIdSuffix ".dev"` so that it can be
+// installed beside the release app -- so a tag derived from the package
+// name is a *different* tag on the one build that can be inspected, and
+// the filter that is supposed to show these lines would hide them on
+// exactly the build used to look for them.
+const Tag = "yellowjacket"
+
+// Android's priorities, from android/log.h. These are the values
+// __android_log_write takes; android.go asserts at compile time that
+// they still match the header, so a renumbered platform is a build
+// failure here rather than a warning silently logged as an error.
+const (
+	PrioVerbose = 2
+	PrioDebug   = 3
+	PrioInfo    = 4
+	PrioWarn    = 5
+	PrioError   = 6
+	PrioFatal   = 7
+)
+
+// maxPayload is how much of one line liblog will carry.
+//
+// The kernel logger's entry is 4068 bytes for the tag, the message and
+// their two NULs together, and what does not fit is **dropped without
+// comment** -- so a long line would be truncated in the middle of the
+// thing worth reading. 3500 leaves room for the tag and for the "(N/M)"
+// a continuation carries.
+const maxPayload = 3500
+
+// WriteFunc is the platform sink: one already-formatted line, at one
+// priority, under one tag.
+//
+// It is a parameter rather than a package-level function so that the
+// handler can be driven by a test on a machine with no liblog at all.
+type WriteFunc func(prio int, tag, msg string)
+
+// Priority maps a slog level onto an Android one.
+//
+// slog's levels are open -- a caller may define its own at any int --
+// so this is a banding rather than a lookup: anything below Info is
+// debug, anything at or above Error is error. A custom level between
+// two of the standard ones lands in the band beneath it, which is what
+// slog's own level naming does.
+func Priority(level slog.Level) int {
+	switch {
+	case level < slog.LevelDebug:
+		return PrioVerbose
+	case level < slog.LevelInfo:
+		return PrioDebug
+	case level < slog.LevelWarn:
+		return PrioInfo
+	case level < slog.LevelError:
+		return PrioWarn
+	default:
+		return PrioError
+	}
+}
+
+// Handler formats records with slog's own TextHandler and hands each
+// line to a WriteFunc.
+//
+// It delegates the formatting rather than doing it, because WithAttrs
+// and WithGroup are the half of slog.Handler that is easy to get subtly
+// wrong -- and a logger whose groups are wrong is a logger nobody reads.
+// What it does own is what logcat needs and TextHandler does not know
+// about: the priority, and the fact that a line has a maximum length.
+type Handler struct {
+	write WriteFunc
+
+	// mu guards buf, which the delegate writes into. slog.Handler is
+	// documented as safe for concurrent use.
+	mu       *sync.Mutex
+	buf      *bytes.Buffer
+	delegate slog.Handler
+}
+
+// NewHandler builds a handler over an arbitrary sink.
+//
+// The time and the level are dropped from the formatted line: logcat
+// stamps every entry with both, and repeating them costs a quarter of
+// the width of a phone-sized terminal to say the same thing twice.
+func NewHandler(opts *slog.HandlerOptions, write WriteFunc) *Handler {
+	buf := &bytes.Buffer{}
+
+	inner := &slog.HandlerOptions{}
+	if opts != nil {
+		*inner = *opts
+	}
+
+	user := inner.ReplaceAttr
+	inner.ReplaceAttr = func(groups []string, a slog.Attr) slog.Attr {
+		if len(groups) == 0 && isBuiltin(a) {
+			return slog.Attr{}
+		}
+
+		if user != nil {
+			return user(groups, a)
+		}
+
+		return a
+	}
+
+	return &Handler{
+		write:    write,
+		mu:       &sync.Mutex{},
+		buf:      buf,
+		delegate: slog.NewTextHandler(buf, inner),
+	}
+}
+
+// isBuiltin reports whether an attr is slog's own time or level,
+// rather than a caller's attribute that happens to share the name.
+//
+// ReplaceAttr cannot tell those apart by key. It is called with an
+// empty group path for the built-ins *and* for every top-level
+// attribute, so a key comparison alone silently eats a caller's own
+// "level" or "time" -- which is not hypothetical: the probe that
+// verified this package on the device logged one, and the attribute
+// vanished. The kinds are what separate them, because slog builds the
+// built-ins as slog.Any(LevelKey, r.Level) and slog.Time(TimeKey, ...)
+// and an attribute value of type slog.Level is not something a caller
+// passes by accident.
+func isBuiltin(a slog.Attr) bool {
+	switch a.Key {
+	case slog.TimeKey:
+		return a.Value.Kind() == slog.KindTime
+	case slog.LevelKey:
+		_, ok := a.Value.Any().(slog.Level)
+
+		return ok
+	default:
+		return false
+	}
+}
+
+// Enabled reports whether the level is worth formatting.
+func (h *Handler) Enabled(ctx context.Context, level slog.Level) bool {
+	return h.delegate.Enabled(ctx, level)
+}
+
+// Handle formats one record and writes it out, in as many entries as
+// its length demands.
+func (h *Handler) Handle(ctx context.Context, rec slog.Record) error {
+	h.mu.Lock()
+	defer h.mu.Unlock()
+
+	h.buf.Reset()
+
+	if err := h.delegate.Handle(ctx, rec); err != nil {
+		return err
+	}
+
+	prio := Priority(rec.Level)
+	for _, line := range Chunk(strings.TrimRight(h.buf.String(), "\n")) {
+		h.write(prio, Tag, line)
+	}
+
+	return nil
+}
+
+// WithAttrs returns a handler carrying the given attributes.
+func (h *Handler) WithAttrs(attrs []slog.Attr) slog.Handler {
+	return h.derive(h.delegate.WithAttrs(attrs))
+}
+
+// WithGroup returns a handler that qualifies subsequent attributes.
+func (h *Handler) WithGroup(name string) slog.Handler {
+	return h.derive(h.delegate.WithGroup(name))
+}
+
+// derive shares the buffer and its mutex with the parent.
+//
+// They must be shared rather than copied: the delegate returned by
+// WithAttrs writes into the *same* buffer this one does, so a second
+// mutex would guard nothing and two loggers derived from one would
+// interleave their bytes into a single line.
+func (h *Handler) derive(delegate slog.Handler) *Handler {
+	return &Handler{
+		write:    h.write,
+		mu:       h.mu,
+		buf:      h.buf,
+		delegate: delegate,
+	}
+}
+
+// Chunk splits a formatted record into entries liblog will carry
+// whole.
+//
+// A record short enough to fit is returned as it is, which is nearly
+// every record; the numbering only appears where something was going
+// to be silently truncated anyway. It splits on bytes rather than runes
+// because the limit is a byte count -- a multi-byte rune straddling the
+// boundary is a mojibake character in a log line, against a lost one.
+func Chunk(msg string) []string {
+	if len(msg) <= maxPayload {
+		return []string{msg}
+	}
+
+	var parts []string
+
+	for rest := msg; rest != ""; {
+		n := min(maxPayload, len(rest))
+		parts = append(parts, rest[:n])
+		rest = rest[n:]
+	}
+
+	numbered := make([]string, 0, len(parts))
+	for i, p := range parts {
+		numbered = append(
+			numbered,
+			"("+strconv.Itoa(i+1)+"/"+strconv.Itoa(len(parts))+") "+p,
+		)
+	}
+
+	return numbered
+}

backend/androidlog/androidlog_test.go

@@ -0,0 +1,355 @@
+package androidlog_test
+
+import (
+	"log/slog"
+	"strings"
+	"sync"
+	"testing"
+
+	"yellowjacket/backend/androidlog"
+)
+
+// entry is one call to the sink.
+type entry struct {
+	prio int
+	tag  string
+	msg  string
+}
+
+// recorder is the platform write, on a machine with no platform.
+type recorder struct {
+	mu      sync.Mutex
+	entries []entry
+}
+
+func (r *recorder) write(prio int, tag, msg string) {
+	r.mu.Lock()
+	defer r.mu.Unlock()
+
+	r.entries = append(r.entries, entry{prio: prio, tag: tag, msg: msg})
+}
+
+func (r *recorder) only(t *testing.T) entry {
+	t.Helper()
+
+	r.mu.Lock()
+	defer r.mu.Unlock()
+
+	if len(r.entries) != 1 {
+		t.Fatalf("want exactly one entry, got %d: %v", len(r.entries), r.entries)
+	}
+
+	return r.entries[0]
+}
+
+func newLogger(r *recorder, level slog.Level) *slog.Logger {
+	return slog.New(androidlog.NewHandler(
+		&slog.HandlerOptions{Level: level},
+		r.write,
+	))
+}
+
+// TestPriorityMapsEveryLevel pins the level banding.
+//
+// This is the one thing in #160 that a wrong answer hides rather than
+// breaks: logcat prints whatever priority it is handed, so an Error
+// filed as Info is a line that is present, correct and invisible to
+// every filter anyone would use to look for it.
+func TestPriorityMapsEveryLevel(t *testing.T) {
+	t.Parallel()
+
+	tests := []struct {
+		name  string
+		level slog.Level
+		want  int
+	}{
+		{"below debug is verbose", slog.LevelDebug - 1, androidlog.PrioVerbose},
+		{"debug", slog.LevelDebug, androidlog.PrioDebug},
+		{"info", slog.LevelInfo, androidlog.PrioInfo},
+		{"warn", slog.LevelWarn, androidlog.PrioWarn},
+		{"error", slog.LevelError, androidlog.PrioError},
+
+		// slog's levels are open, so a caller may sit between two of
+		// the named ones.  Each lands in the band beneath it, which is
+		// what slog's own level naming does ("INFO+2").
+		{"between info and warn", slog.LevelInfo + 2, androidlog.PrioInfo},
+		{"between warn and error", slog.LevelWarn + 1, androidlog.PrioWarn},
+		{"above error", slog.LevelError + 4, androidlog.PrioError},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			t.Parallel()
+
+			if got := androidlog.Priority(tt.level); got != tt.want {
+				t.Errorf("Priority(%v) = %d, want %d", tt.level, got, tt.want)
+			}
+		})
+	}
+}
+
+// TestPrioritiesAreTheHeadersValues pins the constants themselves.
+//
+// android.go asserts these against android/log.h at compile time, but
+// only a cross-compiler ever builds that file.  This is the assertion
+// that runs in CI, and the numbers are written out longhand on purpose
+// -- comparing a constant to itself would pass on any renumbering.
+func TestPrioritiesAreTheHeadersValues(t *testing.T) {
+	t.Parallel()
+
+	for _, tt := range []struct {
+		name string
+		got  int
+		want int
+	}{
+		{"verbose", androidlog.PrioVerbose, 2},
+		{"debug", androidlog.PrioDebug, 3},
+		{"info", androidlog.PrioInfo, 4},
+		{"warn", androidlog.PrioWarn, 5},
+		{"error", androidlog.PrioError, 6},
+		{"fatal", androidlog.PrioFatal, 7},
+	} {
+		if tt.got != tt.want {
+			t.Errorf("%s priority = %d, want %d", tt.name, tt.got, tt.want)
+		}
+	}
+}
+
+// TestRecordReachesTheSink is the whole point of the package: a line
+// written with slog arrives, under the app's tag, at the right
+// priority.
+func TestRecordReachesTheSink(t *testing.T) {
+	t.Parallel()
+
+	rec := &recorder{}
+	newLogger(rec, slog.LevelInfo).Error("application error", "err", "boom")
+
+	got := rec.only(t)
+
+	if got.prio != androidlog.PrioError {
+		t.Errorf("priority = %d, want %d", got.prio, androidlog.PrioError)
+	}
+
+	if got.tag != androidlog.Tag {
+		t.Errorf("tag = %q, want %q", got.tag, androidlog.Tag)
+	}
+
+	if !strings.Contains(got.msg, "application error") {
+		t.Errorf("message %q does not carry the message", got.msg)
+	}
+
+	if !strings.Contains(got.msg, `err=boom`) {
+		t.Errorf("message %q does not carry the attribute", got.msg)
+	}
+}
+
+// TestTheTagIsNotTheApplicationID guards the trap the tag exists to
+// avoid.
+//
+// The debug build carries `applicationIdSuffix ".dev"`, so it is
+// installed as app.yellowjacket.dev -- and it is the *only* build whose
+// WebView can be inspected, so it is the build anyone debugging this
+// app is running.  A tag derived from the application id therefore
+// differs between the build being looked at and the build the filter
+// was written for, which is the failure this whole issue is about
+// wearing a different hat.
+func TestTheTagIsNotTheApplicationID(t *testing.T) {
+	t.Parallel()
+
+	if strings.Contains(androidlog.Tag, ".") {
+		t.Errorf(
+			"tag %q looks like an application id; it must be stable "+
+				"across the debug suffix",
+			androidlog.Tag,
+		)
+	}
+
+	// Logcat's tag field is 23 bytes.  A longer one is truncated, and a
+	// truncated tag matches no filter.
+	if len(androidlog.Tag) > 23 {
+		t.Errorf("tag %q is %d bytes, over logcat's 23", androidlog.Tag, len(androidlog.Tag))
+	}
+}
+
+// TestTimeAndLevelAreDropped checks the formatting decision.
+//
+// logcat stamps every entry with a timestamp and a priority letter, so
+// carrying slog's own is the same information twice on a 424px screen.
+func TestTimeAndLevelAreDropped(t *testing.T) {
+	t.Parallel()
+
+	rec := &recorder{}
+	newLogger(rec, slog.LevelInfo).Warn("scan finished", "files", 1577)
+
+	got := rec.only(t).msg
+
+	if strings.Contains(got, "time=") {
+		t.Errorf("message %q still carries a timestamp", got)
+	}
+
+	if strings.Contains(got, "level=") {
+		t.Errorf("message %q still carries a level", got)
+	}
+
+	if !strings.Contains(got, "files=1577") {
+		t.Errorf("message %q lost its attributes with them", got)
+	}
+}
+
+// TestACallersOwnLevelAttrSurvives is a regression, and it was found on
+// the phone rather than here.
+//
+// Dropping slog's built-in time and level by key alone also drops a
+// caller's attribute of the same name, because ReplaceAttr sees an
+// empty group path for both. The probe that verified this package on
+// the device wrote slog.Info("...", "level", "info") and logcat showed
+// the message with no attributes at all.
+func TestACallersOwnLevelAttrSurvives(t *testing.T) {
+	t.Parallel()
+
+	rec := &recorder{}
+	newLogger(rec, slog.LevelInfo).Info("probe", "level", "info", "time", "soon")
+
+	got := rec.only(t).msg
+
+	for _, want := range []string{"level=info", "time=soon"} {
+		if !strings.Contains(got, want) {
+			t.Errorf("message %q lost the caller's %q", got, want)
+		}
+	}
+
+	// And slog's own are still gone: the built-in level renders as a
+	// bare word like INFO, never as the caller's value.
+	if strings.Contains(got, "level=INFO") {
+		t.Errorf("message %q carries slog's own level", got)
+	}
+}
+
+// TestLevelIsHonoured checks that Enabled reaches the delegate.
+func TestLevelIsHonoured(t *testing.T) {
+	t.Parallel()
+
+	rec := &recorder{}
+	log := newLogger(rec, slog.LevelWarn)
+
+	log.Info("not this one")
+	log.Warn("this one")
+
+	if got := rec.only(t).msg; !strings.Contains(got, "this one") {
+		t.Errorf("wrong record survived: %q", got)
+	}
+}
+
+// TestGroupsAndAttrsSurvive covers the half of slog.Handler this
+// delegates rather than implements -- the reason it delegates at all.
+func TestGroupsAndAttrsSurvive(t *testing.T) {
+	t.Parallel()
+
+	rec := &recorder{}
+	log := newLogger(rec, slog.LevelInfo).
+		With("component", "player").
+		WithGroup("track")
+
+	log.Info("loaded", "path", "/sdcard/Music/a.flac")
+
+	got := rec.only(t).msg
+
+	for _, want := range []string{
+		"component=player",
+		"track.path=/sdcard/Music/a.flac",
+	} {
+		if !strings.Contains(got, want) {
+			t.Errorf("message %q is missing %q", got, want)
+		}
+	}
+}
+
+// TestDerivedHandlersDoNotInterleave is why derive shares the buffer's
+// mutex rather than taking a new one.
+//
+// Two loggers derived from one write into the same buffer, so a second
+// mutex would guard nothing and a concurrent pair would splice each
+// other's bytes into a single line -- which reads as corrupted logs
+// under load and as nothing at all in a test that logs once.
+func TestDerivedHandlersDoNotInterleave(t *testing.T) {
+	t.Parallel()
+
+	rec := &recorder{}
+	base := newLogger(rec, slog.LevelInfo)
+
+	var wg sync.WaitGroup
+
+	for i := range 8 {
+		wg.Add(1)
+
+		go func() {
+			defer wg.Done()
+
+			log := base.With("worker", i).WithGroup("g")
+			for range 50 {
+				log.Info("tick", "n", i)
+			}
+		}()
+	}
+
+	wg.Wait()
+
+	rec.mu.Lock()
+	defer rec.mu.Unlock()
+
+	if len(rec.entries) != 8*50 {
+		t.Fatalf("got %d entries, want %d", len(rec.entries), 8*50)
+	}
+
+	for _, e := range rec.entries {
+		if strings.Count(e.msg, "msg=tick") != 1 {
+			t.Fatalf("interleaved line: %q", e.msg)
+		}
+	}
+}
+
+// TestChunkLeavesShortLinesAlone is the common case: no numbering
+// appears on a record that was never going to be truncated.
+func TestChunkLeavesShortLinesAlone(t *testing.T) {
+	t.Parallel()
+
+	got := androidlog.Chunk("msg=short")
+
+	if len(got) != 1 || got[0] != "msg=short" {
+		t.Errorf("Chunk(short) = %q, want the input unchanged", got)
+	}
+}
+
+// TestChunkSplitsWhatWouldBeTruncated covers the case liblog drops
+// silently.
+func TestChunkSplitsWhatWouldBeTruncated(t *testing.T) {
+	t.Parallel()
+
+	const n = 9000
+
+	long := strings.Repeat("x", n)
+	parts := androidlog.Chunk(long)
+
+	if len(parts) < 2 {
+		t.Fatalf("a %d-byte line was not split", n)
+	}
+
+	var payload strings.Builder
+
+	for i, p := range parts {
+		if len(p) > 4000 {
+			t.Errorf("part %d is %d bytes, over liblog's entry", i, len(p))
+		}
+
+		_, rest, found := strings.Cut(p, ") ")
+		if !found {
+			t.Fatalf("part %d carries no (n/m) marker: %q", i, p)
+		}
+
+		payload.WriteString(rest)
+	}
+
+	if payload.String() != long {
+		t.Errorf("the parts do not reassemble into the input")
+	}
+}

backend/database/sql/queries/albums.sql

@@ -40,7 +40,8 @@ WHERE id = ? AND (mbid IS NULL OR mbid = '');
 -- name: GetAlbumsWithPendingReleaseMBID :many
 SELECT id, pending_release_mbid FROM albums
 WHERE pending_release_mbid IS NOT NULL AND pending_release_mbid != ''
-  AND (mbid IS NULL OR mbid = '');
+  AND (mbid IS NULL OR mbid = '')
+LIMIT ?;
 
 -- name: DeleteAlbum :exec
 DELETE FROM albums WHERE id = ?;

backend/database/sql/sqlcgen/albums.sql.go

@@ -331,6 +331,7 @@ const getAlbumsWithPendingReleaseMBID = `-- name: GetAlbumsWithPendingReleaseMBI
 SELECT id, pending_release_mbid FROM albums
 WHERE pending_release_mbid IS NOT NULL AND pending_release_mbid != ''
   AND (mbid IS NULL OR mbid = '')
+LIMIT ?
 `
 
 type GetAlbumsWithPendingReleaseMBIDRow struct {
@@ -338,8 +339,8 @@ type GetAlbumsWithPendingReleaseMBIDRow struct {
 	PendingReleaseMbid sql.NullString
 }
 
-func (q *Queries) GetAlbumsWithPendingReleaseMBID(ctx context.Context) ([]GetAlbumsWithPendingReleaseMBIDRow, error) {
-	rows, err := q.db.QueryContext(ctx, getAlbumsWithPendingReleaseMBID)
+func (q *Queries) GetAlbumsWithPendingReleaseMBID(ctx context.Context, limit int64) ([]GetAlbumsWithPendingReleaseMBIDRow, error) {
+	rows, err := q.db.QueryContext(ctx, getAlbumsWithPendingReleaseMBID, limit)
 	if err != nil {
 		return nil, err
 	}

backend/explore/explore.go

@@ -2,6 +2,7 @@ package explore
 
 import (
 	"context"
+	"database/sql"
 	"log/slog"
 	"math"
 	"sort"
@@ -12,6 +13,7 @@ import (
 	"golang.org/x/sync/singleflight"
 
 	"yellowjacket/backend/database"
+	"yellowjacket/backend/database/sql/sqlcgen"
 	"yellowjacket/backend/events"
 	"yellowjacket/backend/jobs"
 )
@@ -279,37 +281,32 @@ func (e *Service) BackfillReleaseGroupMBIDs() {
 	go e.backfillReleaseGroupMBIDs(e.ctx)
 }
 
-func (e *Service) backfillReleaseGroupMBIDs(ctx context.Context) {
-	rows, err := e.db.QueryContext(
-		"SELECT id, pending_release_mbid FROM release_groups "+
-			"WHERE (mbid IS NULL OR mbid = '') "+
-			"AND pending_release_mbid IS NOT NULL AND pending_release_mbid != '' "+
-			"LIMIT ?",
-		releaseGroupMBIDBackfillMaxPerRun,
+// pendingReleaseMBIDs is the albums this pass has work to do on.
+//
+// It is separate from the pass, and returns its error rather than
+// logging it, so that a test can assert the statement runs against the
+// real schema. That is not a general preference -- it is this
+// statement's history: it named `release_groups`, a table plan 013
+// renamed to `albums`, so it failed on every launch since e7748f1 and
+// the pass returned quietly having done nothing. A test of the pass
+// as a whole cannot see that, because a query error and an empty
+// library are the same early return.
+func (e *Service) pendingReleaseMBIDs(
+	ctx context.Context,
+) ([]sqlcgen.GetAlbumsWithPendingReleaseMBIDRow, error) {
+	return e.db.ReadQueries.GetAlbumsWithPendingReleaseMBID(
+		ctx, releaseGroupMBIDBackfillMaxPerRun,
 	)
+}
+
+func (e *Service) backfillReleaseGroupMBIDs(ctx context.Context) {
+	pending, err := e.pendingReleaseMBIDs(ctx)
 	if err != nil {
 		e.logger.Warn("release-group mbid backfill: query failed", "error", err)
 
 		return
 	}
 
-	type pendingRow struct {
-		id          int64
-		releaseMBID string
-	}
-
-	var pending []pendingRow
-
-	for rows.Next() {
-		var p pendingRow
-
-		if err := rows.Scan(&p.id, &p.releaseMBID); err == nil {
-			pending = append(pending, p)
-		}
-	}
-
-	_ = rows.Close()
-
 	if len(pending) == 0 {
 		return
 	}
@@ -335,7 +332,7 @@ func (e *Service) backfillReleaseGroupMBIDs(ctx context.Context) {
 
 		job.progress(i, len(pending))
 
-		release, err := e.mb.LookupRelease(ctx, p.releaseMBID)
+		release, err := e.mb.LookupRelease(ctx, p.PendingReleaseMbid.String)
 		if err != nil || release.ReleaseGroupMBID == "" {
 			// Left alone rather than cleared: LookupRelease caches its
 			// answer (success or a release with no group) for 7 days,
@@ -344,12 +341,19 @@ func (e *Service) backfillReleaseGroupMBIDs(ctx context.Context) {
 			continue
 		}
 
-		_, err = e.db.ExecContext(
-			"UPDATE release_groups SET mbid = ?, pending_release_mbid = NULL "+
-				"WHERE id = ? AND (mbid IS NULL OR mbid = '')",
-			release.ReleaseGroupMBID, p.id,
-		)
-		if err != nil {
+		// The writer, not ReadQueries: an UPDATE issued on the
+		// query-only pool fails at runtime with "attempt to write a
+		// readonly database".
+		if err := e.db.Queries.ResolveAlbumPendingReleaseMBID(
+			ctx,
+			sqlcgen.ResolveAlbumPendingReleaseMBIDParams{
+				Mbid: sql.NullString{
+					String: release.ReleaseGroupMBID,
+					Valid:  true,
+				},
+				ID: p.ID,
+			},
+		); err != nil {
 			e.logger.Warn("release-group mbid backfill: update failed", "error", err)
 		}
 	}

backend/explore/pendingreleasembid_test.go

@@ -0,0 +1,250 @@
+package explore
+
+import (
+	"database/sql"
+	"log/slog"
+	"strconv"
+	"testing"
+
+	"yellowjacket/backend/database"
+	"yellowjacket/backend/database/sql/sqlcgen"
+)
+
+// The release-group MBID backfill queried `release_groups`, a table
+// plan 013 renamed to `albums`, so it failed on its first statement on
+// every launch from e7748f1 until #189 -- and the pass swallowed that,
+// because a query error and an empty library are the same early
+// return. Nothing noticed for two reasons worth keeping in mind:
+//
+//   - the statement was **raw SQL**, so sqlc never read it. Every other
+//     statement in the repo was renamed by the same change because sqlc
+//     reads sql/schemas/ and cannot generate against a table that is not
+//     declared. The two sqlc queries this now calls were written by 013
+//     and left uncalled.
+//   - it needs no network and no fixture library to reproduce. The
+//     failure is at prepare time.
+
+// seedPendingAlbum inserts an album whose files carried a release MBID
+// but no release-group MBID, which is what `library.updateMBIDs`
+// leaves behind for this pass to resolve.
+func seedPendingAlbum(
+	t *testing.T,
+	db *database.DB,
+	name, pendingMBID string,
+) int64 {
+	t.Helper()
+
+	res, err := db.ExecContext(
+		"INSERT INTO albums (name, artist_credit, pending_release_mbid) "+
+			"VALUES (?, ?, ?)",
+		name, "Test Artist", pendingMBID,
+	)
+	if err != nil {
+		t.Fatalf("insert albums row: %v", err)
+	}
+
+	id, err := res.LastInsertId()
+	if err != nil {
+		t.Fatalf("last insert id: %v", err)
+	}
+
+	return id
+}
+
+func newPendingTestService(db *database.DB) *Service {
+	return &Service{db: db, logger: slog.Default()}
+}
+
+// TestPendingReleaseMBIDsRunsAgainstTheRealSchema is the regression.
+//
+// It asserts the statement *runs*, which is the whole of what was
+// broken: against the old raw SQL this returns
+// "no such table: release_groups" rather than a row.
+func TestPendingReleaseMBIDsRunsAgainstTheRealSchema(t *testing.T) {
+	t.Parallel()
+
+	db := database.NewTestDB(t)
+	e := newPendingTestService(db)
+
+	want := seedPendingAlbum(t, db, "Pending Album", "release-mbid-1")
+
+	pending, err := e.pendingReleaseMBIDs(db.Ctx)
+	if err != nil {
+		t.Fatalf("the backfill's query failed: %v", err)
+	}
+
+	if len(pending) != 1 {
+		t.Fatalf("got %d pending albums, want 1", len(pending))
+	}
+
+	if pending[0].ID != want {
+		t.Errorf("got album id %d, want %d", pending[0].ID, want)
+	}
+
+	if got := pending[0].PendingReleaseMbid.String; got != "release-mbid-1" {
+		t.Errorf("got pending mbid %q, want %q", got, "release-mbid-1")
+	}
+}
+
+// TestOnlyUnresolvedAlbumsAreReturned pins the two conditions that make
+// the pass idempotent, since between them they are what stops it doing
+// the same MusicBrainz lookups on every launch forever.
+func TestOnlyUnresolvedAlbumsAreReturned(t *testing.T) {
+	t.Parallel()
+
+	db := database.NewTestDB(t)
+	e := newPendingTestService(db)
+
+	pendingID := seedPendingAlbum(t, db, "Still Pending", "release-mbid-1")
+
+	// Already resolved: it has a real MBID, so there is nothing to
+	// look up even though a marker is still sitting on it.
+	resolved := seedPendingAlbum(t, db, "Already Resolved", "release-mbid-2")
+	if err := db.Queries.SetAlbumMBID(db.Ctx, sqlcgen.SetAlbumMBIDParams{
+		Mbid: sql.NullString{String: "rg-mbid", Valid: true},
+		ID:   resolved,
+	}); err != nil {
+		t.Fatalf("set album mbid: %v", err)
+	}
+
+	// Never had a release MBID to resolve in the first place, which is
+	// most of a library.
+	seedPendingAlbum(t, db, "Nothing Pending", "")
+
+	pending, err := e.pendingReleaseMBIDs(db.Ctx)
+	if err != nil {
+		t.Fatalf("the backfill's query failed: %v", err)
+	}
+
+	if len(pending) != 1 || pending[0].ID != pendingID {
+		t.Fatalf(
+			"got %d albums %v, want only the unresolved one (%d)",
+			len(pending), pending, pendingID,
+		)
+	}
+}
+
+// TestResolvingClearsTheMarker is the other half: once the lookup has
+// answered, the album must stop being a candidate, or the pass repeats
+// the same live MusicBrainz call on every launch.
+func TestResolvingClearsTheMarker(t *testing.T) {
+	t.Parallel()
+
+	db := database.NewTestDB(t)
+	e := newPendingTestService(db)
+
+	id := seedPendingAlbum(t, db, "Pending Album", "release-mbid-1")
+
+	// The writer, deliberately: this is an UPDATE, and the read pool
+	// would refuse it at runtime.
+	if err := db.Queries.ResolveAlbumPendingReleaseMBID(
+		db.Ctx,
+		sqlcgen.ResolveAlbumPendingReleaseMBIDParams{
+			Mbid: sql.NullString{String: "resolved-rg-mbid", Valid: true},
+			ID:   id,
+		},
+	); err != nil {
+		t.Fatalf("resolve pending release mbid: %v", err)
+	}
+
+	pending, err := e.pendingReleaseMBIDs(db.Ctx)
+	if err != nil {
+		t.Fatalf("the backfill's query failed: %v", err)
+	}
+
+	if len(pending) != 0 {
+		t.Fatalf("a resolved album is still a candidate: %v", pending)
+	}
+
+	album, err := db.ReadQueries.GetAlbum(db.Ctx, id)
+	if err != nil {
+		t.Fatalf("get album: %v", err)
+	}
+
+	if album.Mbid.String != "resolved-rg-mbid" {
+		t.Errorf("album mbid = %q, want the resolved one", album.Mbid.String)
+	}
+
+	if album.PendingReleaseMbid.Valid &&
+		album.PendingReleaseMbid.String != "" {
+		t.Errorf(
+			"the pending marker survived as %q",
+			album.PendingReleaseMbid.String,
+		)
+	}
+}
+
+// TestAResolvedMBIDIsNeverOverwritten covers the guard in the UPDATE.
+//
+// The pass runs against rows it read earlier, and a rescan can resolve
+// an album from its tags in between -- a real MBID from the file must
+// win over one this pass inferred from a release.
+func TestAResolvedMBIDIsNeverOverwritten(t *testing.T) {
+	t.Parallel()
+
+	db := database.NewTestDB(t)
+
+	id := seedPendingAlbum(t, db, "Pending Album", "release-mbid-1")
+
+	if err := db.Queries.SetAlbumMBID(db.Ctx, sqlcgen.SetAlbumMBIDParams{
+		Mbid: sql.NullString{String: "from-the-tags", Valid: true},
+		ID:   id,
+	}); err != nil {
+		t.Fatalf("set album mbid: %v", err)
+	}
+
+	if err := db.Queries.ResolveAlbumPendingReleaseMBID(
+		db.Ctx,
+		sqlcgen.ResolveAlbumPendingReleaseMBIDParams{
+			Mbid: sql.NullString{String: "from-the-backfill", Valid: true},
+			ID:   id,
+		},
+	); err != nil {
+		t.Fatalf("resolve pending release mbid: %v", err)
+	}
+
+	album, err := db.ReadQueries.GetAlbum(db.Ctx, id)
+	if err != nil {
+		t.Fatalf("get album: %v", err)
+	}
+
+	if album.Mbid.String != "from-the-tags" {
+		t.Errorf(
+			"album mbid = %q, want the tagged one to have won",
+			album.Mbid.String,
+		)
+	}
+}
+
+// TestThePassIsBounded checks the LIMIT.
+//
+// Each row costs a live MusicBrainz lookup on a 1 req/s limiter shared
+// with every page the user can open, so an unbounded read is a run that
+// lasts as long as the library is untagged. The sqlc query 013 wrote
+// had no LIMIT; the raw statement it was replacing did.
+func TestThePassIsBounded(t *testing.T) {
+	t.Parallel()
+
+	db := database.NewTestDB(t)
+	e := newPendingTestService(db)
+
+	for i := range releaseGroupMBIDBackfillMaxPerRun + 10 {
+		seedPendingAlbum(
+			t, db,
+			"Album "+string(rune('A'+i%26))+strconv.Itoa(i),
+			"release-mbid-"+strconv.Itoa(i),
+		)
+	}
+
+	pending, err := e.pendingReleaseMBIDs(db.Ctx)
+	if err != nil {
+		t.Fatalf("the backfill's query failed: %v", err)
+	}
+
+	if len(pending) != releaseGroupMBIDBackfillMaxPerRun {
+		t.Errorf(
+			"got %d albums, want the run bounded at %d",
+			len(pending), releaseGroupMBIDBackfillMaxPerRun,
+		)
+	}
+}

backend/player/buffered_streamer.go

@@ -47,6 +47,39 @@ type BufferedStreamer struct {
 	// seek bar and suppresses its interpolation.
 	starved      int
 	starvedSince time.Time
+
+	// underruns accumulates for the life of this streamer, where
+	// starved is reset by every arriving sample.
+	//
+	// The two answer different questions and only the first was being
+	// asked. starved is a *stall* detector: it exists to end a track
+	// whose source has died, so it forgets a run the moment audio
+	// resumes -- which is exactly the case this counts. A hundred 20ms
+	// underruns a minute never approach the give-up threshold and were
+	// invisible to the log, the UI and every test tier, while being
+	// audible as static: an underrun is served as a run of zeros
+	// spliced into the waveform, and a step discontinuity at each edge
+	// is what a click is.
+	underruns UnderrunStats
+}
+
+// UnderrunStats is what the ring buffer missed, cumulatively.
+//
+// Samples rather than milliseconds because this type does not know the
+// sample rate -- the player does, and converts at the point of
+// reporting.
+type UnderrunStats struct {
+	// Runs is the number of *episodes*: transitions from healthy into
+	// starved. Calls is how many Stream calls were served with
+	// silence, and Samples is how much silence that was.
+	//
+	// Runs is the count that means something audible. One episode is
+	// one pop however many calls it spans, and the ratio of the two is
+	// how long the average episode was -- which is what separates
+	// "clicking" from "dropping out".
+	Runs    int64
+	Calls   int64
+	Samples int64
 }
 
 // The silence fill is bounded by both a duration and a run of calls,
@@ -226,6 +259,13 @@ func (bs *BufferedStreamer) Stream(
 		// but only for a bounded stretch, because "forever" is
 		// reported upward as healthy playback and there is no watchdog
 		// above this to notice otherwise.
+		if bs.starved == 0 {
+			bs.underruns.Runs++
+		}
+
+		bs.underruns.Calls++
+		bs.underruns.Samples += int64(len(samples))
+
 		bs.starved++
 
 		if bs.starvedSince.IsZero() {
@@ -269,6 +309,20 @@ func (bs *BufferedStreamer) Stream(
 	return n, true
 }
 
+// Underruns returns the cumulative underrun count.
+//
+// It is a snapshot rather than a live view, and it is read from
+// outside the audio callback: counting happens in Stream, under the
+// lock it already takes, because that path has a real-time deadline
+// and anything that allocates or formats on it is a cause of the
+// defect it is measuring rather than a measurement of it.
+func (bs *BufferedStreamer) Underruns() UnderrunStats {
+	bs.mu.Lock()
+	defer bs.mu.Unlock()
+
+	return bs.underruns
+}
+
 // Err returns any error encountered by the source streamer.
 func (bs *BufferedStreamer) Err() error {
 	bs.mu.Lock()
@@ -298,6 +352,10 @@ func (bs *BufferedStreamer) Flush() {
 
 // resetStarvationLocked forgets an underrun run.  Must be called with
 // bs.mu held.
+//
+// Deliberately does not touch bs.underruns: forgetting the run is what
+// makes starved a stall detector, and remembering it is the whole
+// point of the counter beside it.
 func (bs *BufferedStreamer) resetStarvationLocked() {
 	bs.starved = 0
 	bs.starvedSince = time.Time{}

backend/player/buffered_underrun_test.go

@@ -0,0 +1,279 @@
+package player
+
+import (
+	"sync"
+	"testing"
+	"time"
+)
+
+// An underrun is audible and nothing counted it (#135).
+//
+// The distinction these tests exist for is that `starved` and
+// `underruns` disagree on purpose. `starved` is a stall detector: it is
+// reset by every arriving sample, because its job is to end a track
+// whose source has died and a source that is merely slow must not be
+// cut short (TestUnderrunsDoNotAccumulateAcrossASlowSource, next
+// door). That reset is exactly what made the audible case invisible --
+// a hundred short underruns a minute never approach the give-up
+// threshold, and each one is a run of zeros spliced into the waveform
+// with a step discontinuity at both edges.
+
+// TestAnEmptyRingIsCounted is the measurement itself: silence served
+// for a missing sample is recorded rather than merely tolerated.
+func TestAnEmptyRingIsCounted(t *testing.T) {
+	t.Parallel()
+
+	// A source that never produces is the cleanest way to make the
+	// ring empty on demand; the stall budget is far longer than the
+	// handful of calls below.
+	bs := NewBufferedStreamer(stalledStreamer{}, 1024)
+	defer bs.Close()
+
+	if got := bs.Underruns(); got != (UnderrunStats{}) {
+		t.Fatalf("a fresh streamer already reports %+v", got)
+	}
+
+	buf := make([][2]float64, 256)
+
+	for range 3 {
+		if _, ok := bs.Stream(buf); !ok {
+			t.Fatal("the stall budget ran out before the test did")
+		}
+	}
+
+	got := bs.Underruns()
+
+	if got.Calls != 3 {
+		t.Errorf("Calls = %d, want 3", got.Calls)
+	}
+
+	if got.Samples != int64(3*len(buf)) {
+		t.Errorf("Samples = %d, want %d", got.Samples, 3*len(buf))
+	}
+
+	// Three consecutive silent calls are one episode, not three. That
+	// is the number that means something audible: one interruption is
+	// one pop however many callbacks it spans.
+	if got.Runs != 1 {
+		t.Errorf("Runs = %d, want 1 -- an unbroken run is one episode", got.Runs)
+	}
+}
+
+// TestSilenceIsWhatIsCounted pins what an underrun actually does to the
+// waveform, which is the reason to count it at all.
+func TestSilenceIsWhatIsCounted(t *testing.T) {
+	t.Parallel()
+
+	bs := NewBufferedStreamer(stalledStreamer{}, 1024)
+	defer bs.Close()
+
+	buf := make([][2]float64, 64)
+	for i := range buf {
+		buf[i] = [2]float64{0.5, 0.5}
+	}
+
+	n, ok := bs.Stream(buf)
+	if !ok || n != len(buf) {
+		t.Fatalf("Stream = (%d, %v), want (%d, true)", n, ok, len(buf))
+	}
+
+	for i := range buf {
+		if buf[i] != ([2]float64{}) {
+			t.Fatalf("sample %d is %v, want silence", i, buf[i])
+		}
+	}
+
+	if got := bs.Underruns().Samples; got != int64(len(buf)) {
+		t.Errorf("counted %d samples of silence, wrote %d", got, len(buf))
+	}
+}
+
+// TestSeparateEpisodesAreSeparateRuns is the counter's whole shape:
+// audio arriving between two underruns makes them two, because that is
+// two interruptions and two clicks.
+func TestSeparateEpisodesAreSeparateRuns(t *testing.T) {
+	t.Parallel()
+
+	// A source that yields nothing until it is fed, so the ring can be
+	// emptied, filled and emptied again on demand.
+	src := &gatedStreamer{}
+
+	bs := NewBufferedStreamer(src, 1024)
+	defer bs.Close()
+
+	buf := make([][2]float64, 128)
+
+	starve := func() {
+		t.Helper()
+
+		for range 2 {
+			if _, ok := bs.Stream(buf); !ok {
+				t.Fatal("the stall budget ran out before the test did")
+			}
+		}
+	}
+
+	// feed lets exactly one bufferful through and drains it, so the
+	// ring is empty again on return. Allowing more would mean the
+	// starve() after it drained real audio instead of underrunning,
+	// which is what the first version of this test did -- it reported
+	// one episode and looked like the counter was wrong.
+	feed := func() {
+		t.Helper()
+
+		src.allow(len(buf))
+
+		// The read-ahead is a goroutine, so wait for real samples
+		// rather than assuming they have landed.
+		deadline := time.Now().Add(2 * time.Second)
+		for time.Now().Before(deadline) {
+			n, ok := bs.Stream(buf)
+			if ok && n > 0 && buf[0] != ([2]float64{}) {
+				return
+			}
+
+			time.Sleep(time.Millisecond)
+		}
+
+		t.Fatal("the source never delivered a sample")
+	}
+
+	starve()
+	feed()
+	starve()
+
+	if got := bs.Underruns().Runs; got < 2 {
+		t.Errorf(
+			"Runs = %d, want at least 2 -- audio in between makes two "+
+				"episodes, not one",
+			got,
+		)
+	}
+}
+
+// TestTheStallResetDoesNotClearTheCounter is the regression this file
+// is really about.
+//
+// resetStarvationLocked runs on every arriving sample and on every
+// Flush. If it cleared the cumulative count too, the counter would
+// report zero on exactly the workload it exists to measure -- a stream
+// that underruns repeatedly but always recovers -- which is
+// indistinguishable from healthy playback and is what the code did
+// before #135.
+func TestTheStallResetDoesNotClearTheCounter(t *testing.T) {
+	t.Parallel()
+
+	bs := NewBufferedStreamer(stalledStreamer{}, 1024)
+	defer bs.Close()
+
+	buf := make([][2]float64, 128)
+
+	if _, ok := bs.Stream(buf); !ok {
+		t.Fatal("the stall budget ran out before the test did")
+	}
+
+	before := bs.Underruns()
+	if before.Runs == 0 {
+		t.Fatal("nothing was counted, so the reset cannot be tested")
+	}
+
+	// Both of the ways a run is forgotten.
+	bs.mu.Lock()
+	bs.resetStarvationLocked()
+	bs.mu.Unlock()
+
+	bs.Flush()
+
+	if got := bs.Underruns(); got != before {
+		t.Errorf(
+			"forgetting the stall run also discarded the count: %+v, "+
+				"want %+v",
+			got, before,
+		)
+	}
+}
+
+// TestUnderrunDeltaNeverGoesBackwards covers the one arithmetic trap in
+// the reporting side.
+//
+// The counter belongs to the streamer and the streamer is replaced on
+// every track, so a baseline carried across a track change is the
+// previous track's total subtracted from a fresh zero. The load path
+// resets the baseline, and this clamps as well -- a negative count in a
+// log line reads as a broken instrument, which would discredit the
+// measurement rather than merely mis-state it.
+func TestUnderrunDeltaNeverGoesBackwards(t *testing.T) {
+	t.Parallel()
+
+	tests := []struct {
+		name string
+		now  UnderrunStats
+		last UnderrunStats
+		want UnderrunStats
+	}{
+		{
+			name: "ordinary progress",
+			now:  UnderrunStats{Runs: 5, Calls: 40, Samples: 4000},
+			last: UnderrunStats{Runs: 2, Calls: 10, Samples: 1000},
+			want: UnderrunStats{Runs: 3, Calls: 30, Samples: 3000},
+		},
+		{
+			name: "nothing happened",
+			now:  UnderrunStats{Runs: 5, Calls: 40, Samples: 4000},
+			last: UnderrunStats{Runs: 5, Calls: 40, Samples: 4000},
+			want: UnderrunStats{},
+		},
+		{
+			name: "a new streamer, with a stale baseline",
+			now:  UnderrunStats{},
+			last: UnderrunStats{Runs: 9, Calls: 90, Samples: 9000},
+			want: UnderrunStats{},
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			t.Parallel()
+
+			if got := underrunDelta(tt.now, tt.last); got != tt.want {
+				t.Errorf("underrunDelta = %+v, want %+v", got, tt.want)
+			}
+		})
+	}
+}
+
+// gatedStreamer produces only what it has been allowed to, and
+// otherwise stalls without ending -- so a test can decide exactly when
+// the ring runs dry.
+type gatedStreamer struct {
+	mu        sync.Mutex
+	remaining int
+}
+
+func (g *gatedStreamer) allow(n int) {
+	g.mu.Lock()
+	defer g.mu.Unlock()
+
+	g.remaining += n
+}
+
+func (g *gatedStreamer) Stream(samples [][2]float64) (int, bool) {
+	g.mu.Lock()
+	defer g.mu.Unlock()
+
+	if g.remaining <= 0 {
+		return 0, true
+	}
+
+	n := min(len(samples), g.remaining)
+
+	for i := range n {
+		samples[i] = [2]float64{0.25, 0.25}
+	}
+
+	g.remaining -= n
+
+	return n, true
+}
+
+func (g *gatedStreamer) Err() error { return nil }

backend/player/player.go

@@ -39,16 +39,21 @@ type Player struct {
 	// via the queue).
 	mu sync.Mutex
 
-	ctx                     context.Context
-	logger                  *slog.Logger
-	db                      *database.DB
-	state                   State
-	currentFile             *os.File
-	format                  beep.Format
-	baseStreamer            beep.Streamer
-	seeker                  beep.StreamSeeker
-	resampled               beep.Streamer
-	buffered                *BufferedStreamer
+	ctx          context.Context
+	logger       *slog.Logger
+	db           *database.DB
+	state        State
+	currentFile  *os.File
+	format       beep.Format
+	baseStreamer beep.Streamer
+	seeker       beep.StreamSeeker
+	resampled    beep.Streamer
+	buffered     *BufferedStreamer
+
+	// lastUnderruns is the previous report, so the 1 Hz log can say
+	// what happened in the last second and stay quiet when nothing did.
+	// It is reset with the streamer, in loadFileLocked.
+	lastUnderruns           UnderrunStats
 	control                 *beep.Ctrl
 	volume                  *effects.Volume
 	speakerStreamer         beep.Streamer
@@ -292,9 +297,78 @@ func (p *Player) emitPositionIfPlaying() {
 		return
 	}
 
+	p.reportUnderrunsLocked()
 	p.emitPositionLocked()
 }
 
+// underrunDelta is what happened since the last report.
+//
+// It clamps at zero rather than subtracting blind, because the counter
+// belongs to the *streamer* and the streamer is replaced on every
+// track: a baseline carried across that boundary is the previous
+// track's total subtracted from a fresh zero, which is negative. That
+// is repaired at the load (lastUnderruns is reset with the streamer)
+// and clamped here as well, because a negative count in a log line
+// reads as a broken instrument and would discredit the measurement
+// this exists to make.
+func underrunDelta(now, last UnderrunStats) UnderrunStats {
+	return UnderrunStats{
+		Runs:    max(0, now.Runs-last.Runs),
+		Calls:   max(0, now.Calls-last.Calls),
+		Samples: max(0, now.Samples-last.Samples),
+	}
+}
+
+// reportUnderrunsLocked logs what the ring buffer missed, at most once
+// a second and only when the number moved.  Must be called with p.mu
+// held.
+//
+// **An underrun is audible and nothing counted it** (#135). The ring
+// serves silence when it is empty, so a run of zeros is spliced into
+// the waveform and the step discontinuity at each edge is a click; a
+// series of short ones is static. Everything that makes one likelier
+// is worse on a phone than on a desktop -- slower storage, a governor
+// that parks cores, background work, GC -- and no tier here can see it,
+// since CI's audio device is a null sink chosen because it keeps time.
+//
+// Three things about the reporting are deliberate.
+//
+// **It is on the 1 Hz position ticker rather than in Stream.** Stream
+// runs on the speaker callback's real-time deadline, and a log line
+// there would allocate, format and write on the exact path whose
+// missed deadline is the defect -- measuring by making it worse.
+//
+// **An unchanged count is not logged.** That is emitStatus' rule one
+// package over: a healthy player is silent, so anything in the log is
+// news, and the line appears exactly while it is popping. Reading it
+// off a device means `make android-logs` with the audio audible.
+//
+// **It is Info rather than Debug**, because the default level is Info
+// and a phone has no convenient way to set YJ_LOG_LEVEL -- a debug
+// line here would be a counter nobody on the affected platform can
+// read, which is the shape of the bug that made #160 necessary.
+func (p *Player) reportUnderrunsLocked() {
+	if p.buffered == nil {
+		return
+	}
+
+	stats := p.buffered.Underruns()
+	if stats == p.lastUnderruns {
+		return
+	}
+
+	since := underrunDelta(stats, p.lastUnderruns)
+	p.lastUnderruns = stats
+
+	slog.Info("audio underrun",
+		"runs", since.Runs,
+		"calls", since.Calls,
+		"silenceMs", speakerSampleRate.D(int(since.Samples)).Milliseconds(),
+		"trackRuns", stats.Runs,
+		"trackSilenceMs", speakerSampleRate.D(int(stats.Samples)).Milliseconds(),
+	)
+}
+
 // emitPositionLocked pushes the current position to the frontend.
 // Must be called with p.mu held.
 func (p *Player) emitPositionLocked() {
@@ -484,6 +558,12 @@ func (p *Player) updateStreamers(
 		p.resampled, int(speakerSampleRate)*2,
 	)
 
+	// The counter belongs to the streamer, so the baseline it is
+	// reported against has to go with it -- otherwise the first report
+	// of a new track is the previous track's total subtracted from
+	// zero, which is negative and looks like the instrument is broken.
+	p.lastUnderruns = UnderrunStats{}
+
 	// wrap in ctrl streamer to allow play/pause
 	p.control = &beep.Ctrl{Streamer: p.buffered}
 

backend/system/userdata.go

@@ -52,6 +52,75 @@ func UseHomeOverride(base string) {
 	_ = os.Setenv(envHomeOverride, base)
 }
 
+// envTempDir is the variable Go's os.TempDir() reads, and through it
+// every library in the process that asks for a temporary file.
+const envTempDir = "TMPDIR"
+
+// tempDirName is the subdirectory of the app's own storage that
+// becomes that answer.
+const tempDirName = "tmp"
+
+// UseTempDir gives the process a temporary directory that exists.
+//
+// **Android has no /tmp and hands an app no TMPDIR**, and Go's
+// os.TempDir() falls back to "/tmp" when the variable is unset -- so
+// every library in the process that wants scratch space is handed a
+// path that has never existed. SQLite is the one that noticed: an
+// INSERT ... SELECT large enough to spill returned
+// SQLITE_IOERR_GETTEMPPATH (disk I/O error 6410), which is how the
+// champion search index came to fail its rebuild on every launch while
+// the app otherwise looked healthy (#190).
+//
+// It is the *class* that is fixed here rather than that statement.
+// Anything that spills fails the same way on that platform -- large
+// sorts, large joins, VACUUM -- so the repair belongs at the process's
+// one answer to the question rather than at each caller. The
+// alternative considered was PRAGMA temp_store = MEMORY, which is
+// cheaper and more local and is a promise that every future spill fits
+// in RAM on a phone; the catalog is the largest thing in this app and
+// that is not a promise worth making silently.
+//
+// The rules are UseHomeOverride's, for the same reasons. **An empty
+// base is a no-op**, because that is what
+// application.Mobile.StoragePath() returns on desktop -- so this needs
+// no build tag and changes nothing off mobile, where /tmp is real. And
+// **an explicit TMPDIR wins**, so anyone who set one deliberately gets
+// what they asked for; nothing sets it on the platform this exists for.
+//
+// It returns its error rather than swallowing it because a temp
+// directory that could not be created is the same silent failure one
+// step earlier, and since #160 a log line on that platform is
+// something a person can actually read.
+func UseTempDir(base string) error {
+	if base == "" || os.Getenv(envTempDir) != "" {
+		return nil
+	}
+
+	dir := filepath.Join(base, tempDirName)
+
+	if err := os.MkdirAll(dir, os.ModePerm); err != nil {
+		return fmt.Errorf("could not make the temp directory %s: %w", dir, err)
+	}
+
+	// Writability is checked rather than assumed: the whole failure
+	// this repairs is a directory that is named and cannot be used, and
+	// MkdirAll on an existing unwritable directory succeeds.
+	probe, err := os.CreateTemp(dir, "probe")
+	if err != nil {
+		return fmt.Errorf("temp directory %s is not writable: %w", dir, err)
+	}
+
+	name := probe.Name()
+	_ = probe.Close()
+	_ = os.Remove(name)
+
+	if err := os.Setenv(envTempDir, dir); err != nil {
+		return fmt.Errorf("could not set %s: %w", envTempDir, err)
+	}
+
+	return nil
+}
+
 // getUserDirPath returns and creates the path for a user directory.
 func getUserDirPath(dt dirType) (string, error) {
 	path, err := resolveUserDirPath(dt)

backend/system/userdata_test.go

@@ -87,3 +87,116 @@ func TestUseHomeOverride(t *testing.T) {
 		})
 	}
 }
+
+// UseTempDir carries UseHomeOverride's two rules for the same reasons,
+// plus one of its own: the directory it names has to be usable.
+//
+// **The only tier that can compile the platform this exists for is a
+// phone**, so everything decidable off one is decided here -- which is
+// androidpayload.go's discipline, and is why the platform call is a
+// parameter rather than something this package reaches for. The
+// device's half is a single measurement: no /tmp, no TMPDIR (#190).
+func TestUseTempDir(t *testing.T) {
+	t.Run("an empty base is a no-op", func(t *testing.T) {
+		// This is the desktop case in full: StoragePath() answers ""
+		// off mobile, where /tmp is real and must be left alone.
+		t.Setenv(envTempDir, "")
+
+		if err := UseTempDir(""); err != nil {
+			t.Fatalf("UseTempDir(\"\") = %v, want nil", err)
+		}
+
+		if got := os.Getenv(envTempDir); got != "" {
+			t.Errorf("%s = %q, want it untouched", envTempDir, got)
+		}
+	})
+
+	t.Run("an explicit TMPDIR wins", func(t *testing.T) {
+		const chosen = "/somewhere/deliberate"
+
+		// The base is taken before TMPDIR moves, because t.TempDir()
+		// reads TMPDIR too -- which is the same fact this function is
+		// about, met from the other side.
+		base := t.TempDir()
+
+		t.Setenv(envTempDir, chosen)
+
+		if err := UseTempDir(base); err != nil {
+			t.Fatalf("UseTempDir = %v, want nil", err)
+		}
+
+		if got := os.Getenv(envTempDir); got != chosen {
+			t.Errorf("%s = %q, want the explicit %q", envTempDir, got, chosen)
+		}
+	})
+
+	t.Run("points at a real directory under the base", func(t *testing.T) {
+		base := t.TempDir()
+
+		t.Setenv(envTempDir, "")
+
+		if err := UseTempDir(base); err != nil {
+			t.Fatalf("UseTempDir = %v, want nil", err)
+		}
+
+		got := os.Getenv(envTempDir)
+
+		want := filepath.Join(base, tempDirName)
+		if got != want {
+			t.Fatalf("%s = %q, want %q", envTempDir, got, want)
+		}
+
+		// The whole failure being repaired is a temp directory that is
+		// named and does not exist, so naming one is not enough.
+		info, err := os.Stat(got)
+		if err != nil {
+			t.Fatalf("the temp directory was named but not created: %v", err)
+		}
+
+		if !info.IsDir() {
+			t.Fatalf("%s is not a directory", got)
+		}
+	})
+
+	t.Run("os.TempDir then answers with it", func(t *testing.T) {
+		// The point of setting the variable at all: this is what every
+		// library in the process reads, SQLite's driver included.
+		base := t.TempDir()
+
+		t.Setenv(envTempDir, "")
+
+		if err := UseTempDir(base); err != nil {
+			t.Fatalf("UseTempDir = %v, want nil", err)
+		}
+
+		if got := os.TempDir(); got != filepath.Join(base, tempDirName) {
+			t.Errorf("os.TempDir() = %q, want the directory we made", got)
+		}
+	})
+
+	t.Run("an unwritable directory is an error, not a silent success", func(t *testing.T) {
+		if os.Getuid() == 0 {
+			t.Skip("root can write anywhere, so there is nothing to refuse")
+		}
+
+		base := t.TempDir()
+
+		// MkdirAll on an existing directory succeeds whatever its
+		// mode, so without the write probe this case would set TMPDIR
+		// to a directory nothing can use -- which is the bug again,
+		// one directory over.
+		if err := os.Mkdir(filepath.Join(base, tempDirName), 0o500); err != nil {
+			t.Fatalf("prepare the unwritable directory: %v", err)
+		}
+
+		t.Setenv(envTempDir, "")
+
+		if err := UseTempDir(base); err == nil {
+			t.Fatal("UseTempDir accepted a directory it cannot write to")
+		}
+
+		if got := os.Getenv(envTempDir); got != "" {
+			t.Errorf("%s was set to %q despite the failure", envTempDir, got)
+		}
+	})
+}

e2e/specs/long-press.spec.ts

@@ -1,127 +0,0 @@
-import { test, expect } from '../support/fixtures.js';
-
-/**
- * Long-press is the touch route to a context menu (plan 016 B2 phase 3).
- *
- * The component tier proves the gesture in isolation, against markup it
- * built itself. What it cannot prove is the half that made this one
- * listener instead of six: that the synthetic event reaches the handler
- * a *real* component bound — `track-list` delegates its `contextmenu`
- * on the `lit-virtualizer` rather than binding one per row — and that
- * the real `wa-popup` menu opens from it, which is a path with its own
- * history of opening and then refusing to work (see
- * `menu-keyboard.spec.ts`).
- *
- * The pointer events are dispatched rather than performed: this project
- * runs Desktop Chrome and Desktop Safari, neither of which has touch,
- * and a device tier does not exist. So this is honest about what it
- * checks — the app's own listeners, on the app's own DOM, from the
- * events a touch would produce — and not about a real finger.
- */
-
-/** A common small phone, as in `phone-shell.spec.ts`. */
-const PHONE = { width: 390, height: 844 };
-
-/** Comfortably past the module's 500ms hold. */
-const HELD = 900;
-
-type Page = import('@playwright/test').Page;
-
-/** The track list's menu panel, or null while it is not rendered. */
-const panel = (page: Page) =>
-  page.evaluate(() => {
-    const el = document
-      .querySelector('track-list')
-      ?.shadowRoot?.querySelector('.context-menu-panel');
-
-    if (!el) return null;
-
-    return {
-      role: el.getAttribute('role'),
-      label: el.getAttribute('aria-label'),
-      items: el.querySelectorAll('[role="menuitem"]').length,
-    };
-  });
-
-/**
- * Press the first track row, optionally dragging partway through — the
- * shape of a scroll that begins on a row, which must not open a menu.
- */
-async function pressFirstRow(
-  page: Page,
-  opts: { driftY?: number } = {},
-): Promise<void> {
-  await page.evaluate((drift) => {
-    // `.track-row`, not `[role="row"]`: the column header is a row too,
-    // and it is the *first* one -- a press on it is correctly ignored,
-    // which reads exactly like the gesture not working.
-    const row = document
-      .querySelector('track-list')
-      ?.shadowRoot?.querySelector('.track-row');
-
-    if (!row) throw new Error('no track row to press');
-
-    const box = row.getBoundingClientRect();
-    const x = Math.round(box.left + box.width / 2);
-    const y = Math.round(box.top + box.height / 2);
-    const send = (type: string, dy = 0) =>
-      row.dispatchEvent(
-        new PointerEvent(type, {
-          bubbles: true,
-          composed: true,
-          cancelable: true,
-          pointerType: 'touch',
-          isPrimary: true,
-          clientX: x,
-          clientY: y + dy,
-        }),
-      );
-
-    send('pointerdown');
-
-    if (drift) send('pointermove', drift);
-  }, opts.driftY ?? 0);
-}
-
-test.describe('long-press opens the track menu', () => {
-  test.beforeEach(async ({ app }) => {
-    await app.setViewportSize(PHONE);
-    await app.getByTestId('tab-tracks').click();
-    await expect(app.getByTestId('main-content')).toHaveAttribute(
-      'data-active-view',
-      'tracks',
-    );
-  });
-
-  test.afterEach(async ({ app }) => {
-    // Every other spec file runs against a desktop, and the viewport
-    // belongs to the shared context rather than to this file.
-    await app.setViewportSize({ width: 1440, height: 900 });
-  });
-
-  test('reaches the delegated handler and opens the real menu', async ({
-    app,
-  }) => {
-    await expect.poll(() => panel(app)).toBeNull();
-
-    await pressFirstRow(app);
-
-    await expect
-      .poll(() => panel(app), { timeout: HELD + 2000 })
-      .toMatchObject({ role: 'menu', label: 'Track actions' });
-
-    // The same panel Shift+F10 opens, items and all -- not an empty
-    // popup that happened to become visible.
-    expect((await panel(app))?.items).toBeGreaterThan(0);
-  });
-
-  test('does not open one for a press that turns into a scroll', async ({
-    app,
-  }) => {
-    await pressFirstRow(app, { driftY: 40 });
-
-    await app.waitForTimeout(HELD);
-
-    expect(await panel(app)).toBeNull();
-  });
-});

e2e/specs/phone-now-playing.spec.ts

@@ -0,0 +1,283 @@
+import { test, expect } from '../support/fixtures.js';
+
+/**
+ * Now Playing on a short screen (#51).
+ *
+ * #51 asks for a layout that "survives" ~424x439 with the controls
+ * never scrolling off. #172 measured why it did not — the stacked
+ * layout's budget is fixed, so the art gets whatever is left, and that
+ * was 39px before #64 and 53px after it.
+ *
+ * **Two separate claims are asserted here, and only one of them is
+ * about the phone.**
+ *
+ * The first is that the art is *square*. It was not: `aspect-ratio` is
+ * specified not to re-derive the width when `max-height` clamps the
+ * height, so the art was drawn as a letterbox band and `object-fit:
+ * cover` cropped the cover to it — 264x53 on the reference device. The
+ * leftover only exceeds the width above ~843px of viewport, so this
+ * was every height from ~500 to ~843 as well: most phones, and any
+ * short window. That is ordinary CSS rather than a Chrome 113 quirk,
+ * so this tier can see it, and the heights below are chosen to cover
+ * the range rather than the one device.
+ *
+ * The second is the reflow: below 500px the art and the names sit side
+ * by side, which is what takes the art from 53px to 143px. That is
+ * asserted as a *relation between boxes* — the art beside the names,
+ * not above them — because the pixel count is a consequence of the
+ * arrangement and would pin this file to one device's chrome.
+ *
+ * **What this tier cannot see** is the device's engine: CI's Chromium
+ * and WebKit are current, and #60's clipping showed what that costs.
+ * Nothing here depends on Chrome 113 behaviour — the sizing rules were
+ * checked against the device itself, at column heights of 288, 300,
+ * 451, 600 and 800, and the numbers are on #51.
+ */
+type Page = import('@playwright/test').Page;
+
+/** The reference device's real viewport. */
+const DEVICE = { width: 424, height: 439 };
+
+/**
+ * A tall phone, above the reflow's 500px. Roughly a Pixel 7, which is
+ * #51's other named device and was not attached — so what is checked
+ * here is the layout it *should* get, not that device.
+ */
+const TALL_PHONE = { width: 412, height: 869 };
+
+/** Inside the crop's old range and above the reflow: a short window. */
+const SHORT_WINDOW = { width: 390, height: 700 };
+
+/**
+ * The height the layout reflows at. Written down once here because the
+ * specs have to know which arrangement to *wait* for, not only which
+ * to assert.
+ */
+const REFLOW_AT = 500;
+
+/** Put a track in the player, so the view has art and names to lay out. */
+async function stageATrack(page: Page): Promise<void> {
+  await page.evaluate(async () => {
+    const tracks = (await window.__yjEvents.call(
+      'library.Library.GetTracks',
+      [0],
+      10_000,
+    )) as { FilePath: string }[];
+
+    await window.__yjEvents.call(
+      'queue.Queue.SetQueue',
+      [tracks.slice(0, 4).map((t) => t.FilePath), 0, false, { type: '', id: 0, label: '' }],
+      10_000,
+    );
+  });
+}
+
+/** Open the full-screen view and wait for the shell to say so. */
+async function openNowPlaying(page: Page): Promise<void> {
+  await page.evaluate(() => {
+    document.dispatchEvent(
+      new CustomEvent('navigate', {
+        detail: { view: 'now-playing' },
+        bubbles: true,
+      }),
+    );
+  });
+
+  await expect(page.getByTestId('main-content')).toHaveAttribute(
+    'data-active-view',
+    'now-playing',
+  );
+
+  // The attribute is the shell's bookkeeping and lands before the view
+  // has a track, so measuring on it alone races the first layout --
+  // which showed up as a 60x5 art on the first spec of a cold run.
+  //
+  // Waiting for a non-zero box is not enough on its own either: a
+  // previous test leaves the *other* arrangement on screen, and a
+  // stale column satisfies "has a size" perfectly. So the wait is for
+  // the arrangement this viewport should have, which is the thing
+  // every assertion below depends on. Found by this file passing one
+  // test at a time and failing in file order.
+  const wantRow = (page.viewportSize()?.height ?? 0) <= REFLOW_AT;
+
+  await page.waitForFunction(
+    (row: boolean) => {
+      const v = document.querySelector('now-playing-view');
+      const stack = v?.shadowRoot?.querySelector('.stack');
+      const el = v?.shadowRoot?.querySelector('.art img, .art .placeholder');
+      const t = v?.shadowRoot?.querySelector('.transport');
+
+      if (!el || !t) return false;
+
+      // A build with no `.stack` at all is the one before this change,
+      // and the squareness assertions are still meaningful against it
+      // -- so this waits for the arrangement only where there is one to
+      // wait for. Otherwise reverting the component to check that these
+      // tests bite produces eight timeouts instead of the measurements
+      // that make the case.
+      if (stack) {
+        const dir = getComputedStyle(stack).flexDirection;
+
+        if (dir !== (row ? 'row' : 'column')) return false;
+      }
+
+      const r = el.getBoundingClientRect();
+
+      return r.width > 0 && r.height > 0 && t.getBoundingClientRect().height > 0;
+    },
+    wantRow,
+  );
+}
+
+/**
+ * The boxes this file reasons about, read in one evaluate.
+ *
+ * It reaches into the view's shadow root rather than using locators
+ * because the question is geometric — where these boxes are *relative
+ * to each other* — and a testid per edge would be four locators and
+ * four round trips to say one thing.
+ */
+async function boxes(page: Page) {
+  return page.evaluate(() => {
+    const v = document.querySelector('now-playing-view');
+
+    if (!v || !v.shadowRoot) return null;
+
+    const rect = (sel: string) => {
+      const el = v.shadowRoot!.querySelector(sel);
+
+      if (!el) return null;
+
+      const r = el.getBoundingClientRect();
+
+      return {
+        left: r.left, right: r.right, top: r.top, bottom: r.bottom,
+        width: r.width, height: r.height,
+      };
+    };
+
+    return {
+      // Whichever of the two the track has; both carry the sizing.
+      art: rect('.art img') ?? rect('.art .placeholder'),
+      artBox: rect('.art'),
+      stack: rect('.stack'),
+      meta: rect('.meta'),
+      transport: rect('.transport'),
+      scrollHeight: v.scrollHeight,
+      clientHeight: v.clientHeight,
+    };
+  });
+}
+
+test.describe('Now Playing survives a short screen', () => {
+  test.beforeEach(async ({ app }) => {
+    await stageATrack(app);
+  });
+
+  /**
+   * The crop, at four heights spanning the range it covered. This is
+   * the assertion that fails on the build before this change: at
+   * 424x439 the art measured 264x53.
+   */
+  for (const vp of [DEVICE, SHORT_WINDOW, TALL_PHONE, { width: 900, height: 500 }]) {
+    test(`draws the art square at ${vp.width}x${vp.height}`, async ({ app }) => {
+      await app.setViewportSize(vp);
+      await openNowPlaying(app);
+
+      const b = await boxes(app);
+
+      expect(b, 'now-playing-view did not mount').not.toBeNull();
+      expect(b!.art, 'neither art nor placeholder rendered').not.toBeNull();
+
+      const { width, height } = b!.art!;
+
+      expect(width, 'the art has no width').toBeGreaterThan(0);
+      // One pixel of slack for sub-pixel layout, and no more: the
+      // defect this guards was a 5:1 band.
+      expect(
+        Math.abs(width - height),
+        `art is ${Math.round(width)}x${Math.round(height)}, not square`,
+      ).toBeLessThanOrEqual(1);
+    });
+  }
+
+  /**
+   * The promise #51 states and plan 018's matrix repeats. A floor on
+   * the art with the block scrolling was the other option on #172 and
+   * this is why it was not taken.
+   */
+  test('never scrolls the transport off the bottom', async ({ app }) => {
+    await app.setViewportSize(DEVICE);
+    await openNowPlaying(app);
+
+    const b = await boxes(app);
+
+    expect(b!.transport!.bottom).toBeLessThanOrEqual(DEVICE.height);
+    expect(
+      b!.scrollHeight,
+      'the view scrolls, so the transport can be moved off screen',
+    ).toBeLessThanOrEqual(b!.clientHeight + 1);
+  });
+
+  /**
+   * The reflow itself, as a relation rather than a measurement: below
+   * 500px the names are *beside* the art, above it they are below.
+   */
+  test('puts the names beside the art below 500px', async ({ app }) => {
+    await app.setViewportSize(DEVICE);
+    await openNowPlaying(app);
+
+    const b = await boxes(app);
+
+    expect(
+      b!.meta!.left,
+      'the names are not to the right of the art',
+    ).toBeGreaterThanOrEqual(b!.artBox!.right - 1);
+  });
+
+  test('keeps the names below the art on a tall phone', async ({ app }) => {
+    await app.setViewportSize(TALL_PHONE);
+    await openNowPlaying(app);
+
+    const b = await boxes(app);
+
+    expect(
+      b!.meta!.top,
+      'the names are not below the art',
+    ).toBeGreaterThanOrEqual(b!.artBox!.bottom - 1);
+  });
+
+  /**
+   * What the reflow actually does, stated as a mechanism rather than
+   * as a number: in a row the art is bounded by the row's *height*,
+   * so it fills it — where in a column it is the leftover after the
+   * names, which is what made it 53px.
+   *
+   * **The pixel count is deliberately not asserted here.** Two drafts
+   * tried. The first compared the art against the column's leftover
+   * computed from the boxes on screen and passed on the broken build,
+   * because the subtraction goes negative when the names are taller
+   * than the art — precisely the defect. The second put a floor of
+   * 100px on it, passed locally at 114 and **failed in CI at 64**: this
+   * app is long-lived, so a job staged by an earlier spec is still on
+   * screen, and the volume control renders here where it does not on
+   * Android. Both are chrome above and below this view, and both move
+   * the leftover. A test that asserts how much room CI happened to
+   * have is a test about the runner.
+   *
+   * The device numbers — 53px to 143px — are on #51, measured there,
+   * which is the only tier that can honestly produce them.
+   */
+  test('fills the row with the art rather than the leftover', async ({ app }) => {
+    await app.setViewportSize(DEVICE);
+    await openNowPlaying(app);
+
+    const b = await boxes(app);
+
+    expect(b!.stack, 'there is no row to fill').not.toBeNull();
+    expect(
+      Math.abs(b!.artBox!.height - b!.stack!.height),
+      'the art does not fill the row, so it is still a leftover',
+    ).toBeLessThanOrEqual(1);
+  });
+});

e2e/specs/queue-as-a-screen.spec.ts

@@ -179,8 +179,8 @@ test.describe('the queue is a screen where it covers the content', () => {
    */
 
   /**
-   * With the panel spanning the whole width the scrim has no uncovered
-   * pixels, so the close button is the only pointer route out of a
+   * With the panel spanning the whole width there is no scrim here at
+   * all (#171), so the close button is the only pointer route out of a
    * full-screen surface. Measured at 424×439 before #55: **25×21px**.
    */
   test('offers a way out a thumb can hit', async ({ app }) => {
@@ -194,6 +194,33 @@ test.describe('the queue is a screen where it covers the content', () => {
     expect(box!.width).toBeGreaterThanOrEqual(44);
     expect(box!.height).toBeGreaterThanOrEqual(44);
   });
+
+  /**
+   * #171 — and it draws no scrim, because there is nowhere to tap.
+   *
+   * `.panel-content` is `width: 100%` here, so the scrim sat entirely
+   * underneath it: measured at 424×439, host, panel and scrim all
+   * 424×318. #24's tap-outside-to-close cannot exist on a surface with
+   * no outside, and a `cursor: pointer` layer nobody can reach is a
+   * claim the component cannot keep.
+   *
+   * Asserted as absence rather than by clicking, for the reason the
+   * issue gives: a naive phone case clicks the scrim's centre and hits
+   * the panel, so it passes on the build this exists to fail. The scrim
+   * is still real between 600 and 899px, which `queue-overlay.spec.ts`
+   * asserts at 900×600 by clicking it.
+   */
+  test('draws no scrim, because a screen has no outside to tap', async ({
+    app,
+  }) => {
+    await openTheQueue(app);
+
+    const scrim = await queue(app).evaluate(
+      (el) => el.shadowRoot!.querySelector('.scrim') !== null,
+    );
+
+    expect(scrim).toBe(false);
+  });
 });
 
 /**

e2e/specs/touch-gestures.spec.ts

@@ -0,0 +1,194 @@
+import { test, expect } from '../support/fixtures.js';
+
+/**
+ * The touch gestures against the real app (plan 019, #63; long-press
+ * from plan 016 B2).
+ *
+ * The component tier proves the gestures in isolation, against markup
+ * it built itself. What it cannot prove is the half that made this one
+ * document listener instead of six: that the announced gesture reaches
+ * the handler a *real* component bound — `track-list` delegates on the
+ * `lit-virtualizer` rather than binding per row — and that the real
+ * menu opens from it, a path with its own history of opening and then
+ * refusing to work (see `menu-keyboard.spec.ts`).
+ *
+ * **Both halves of the reassignment are here, and the second is the
+ * one that matters.** #63 makes a hold on a *track row* mean selection
+ * mode; every other surface in the app keeps the context menu it has
+ * had, because an unclaimed `yj-long-press` still becomes a
+ * `contextmenu`. A spec that only checked the row would pass on a
+ * build that had silently broken the other thirteen menus.
+ *
+ * The pointer events are dispatched rather than performed: this
+ * project runs Desktop Chrome and Desktop Safari, neither of which has
+ * touch. So this is honest about what it checks — the app's own
+ * listeners, on the app's own DOM, from the events a touch would
+ * produce — and not about a real finger. The finger is the Android
+ * tier, and it found something this cannot see: Chrome 113's WebView
+ * fires its own `contextmenu` on a long press, which is why the module
+ * announces the gesture from a native event rather than standing down.
+ */
+
+/** A common small phone, as in `phone-shell.spec.ts`. */
+const PHONE = { width: 390, height: 844 };
+
+/** Comfortably past the module's 500ms hold. */
+const HELD = 900;
+
+type Page = import('@playwright/test').Page;
+
+/** A component's menu panel, or null while it is not rendered. */
+const panel = (page: Page, host: string) =>
+  page.evaluate((tag) => {
+    const el = document
+      .querySelector(tag)
+      ?.shadowRoot?.querySelector('.context-menu-panel');
+
+    if (!el) return null;
+
+    return {
+      role: el.getAttribute('role'),
+      label: el.getAttribute('aria-label'),
+      items: el.querySelectorAll('[role="menuitem"]').length,
+    };
+  }, host);
+
+/** How many tracks the selection bar says are selected, or null. */
+const selectionCount = (page: Page) =>
+  page.evaluate(() => {
+    const bar = document
+      .querySelector('track-list')
+      ?.shadowRoot?.querySelector('selection-bar');
+
+    return bar ? (bar as unknown as { count: number }).count : null;
+  });
+
+/**
+ * Press an element, optionally dragging partway through — the shape of
+ * a scroll that begins on a row, which must be neither gesture — and
+ * optionally lifting, which is what makes it a tap rather than a hold.
+ */
+async function press(
+  page: Page,
+  selector: { host: string; inner: string },
+  opts: { driftY?: number; lift?: boolean } = {},
+): Promise<void> {
+  await page.evaluate(
+    ({ host, inner, drift, lift }) => {
+      const el = document
+        .querySelector(host)
+        ?.shadowRoot?.querySelector(inner);
+
+      if (!el) throw new Error(`no ${inner} in ${host} to press`);
+
+      const box = el.getBoundingClientRect();
+      const x = Math.round(box.left + box.width / 2);
+      const y = Math.round(box.top + box.height / 2);
+      const send = (type: string, dy = 0) =>
+        el.dispatchEvent(
+          new PointerEvent(type, {
+            bubbles: true,
+            composed: true,
+            cancelable: true,
+            pointerType: 'touch',
+            isPrimary: true,
+            clientX: x,
+            clientY: y + dy,
+          }),
+        );
+
+      send('pointerdown');
+
+      if (drift) send('pointermove', drift);
+      if (lift) send('pointerup');
+    },
+    {
+      host: selector.host,
+      inner: selector.inner,
+      drift: opts.driftY ?? 0,
+      lift: opts.lift ?? false,
+    },
+  );
+}
+
+// `.track-row`, not `[role="row"]`: the column header is a row too, and
+// it is the *first* one — a press on it is correctly ignored, which
+// reads exactly like the gesture not working.
+const TRACK_ROW = { host: 'track-list', inner: '.track-row' };
+
+test.describe('a hold on a track row selects it', () => {
+  test.beforeEach(async ({ app }) => {
+    await app.setViewportSize(PHONE);
+    await app.getByTestId('tab-tracks').click();
+    await expect(app.getByTestId('main-content')).toHaveAttribute(
+      'data-active-view',
+      'tracks',
+    );
+  });
+
+  test.afterEach(async ({ app }) => {
+    // Every other spec file runs against a desktop, and the viewport
+    // belongs to the shared context rather than to this file.
+    await app.setViewportSize({ width: 1440, height: 900 });
+  });
+
+  test('raises the selection bar rather than the context menu', async ({
+    app,
+  }) => {
+    await expect.poll(() => selectionCount(app)).toBeNull();
+
+    await press(app, TRACK_ROW);
+
+    await expect
+      .poll(() => selectionCount(app), { timeout: HELD + 2000 })
+      .toBe(1);
+
+    // The gesture is claimed, so the menu this hold used to open must
+    // not also be up -- on a phone that would be a sheet over the bar.
+    expect(await panel(app, 'track-list')).toBeNull();
+  });
+
+  test('is neither gesture when the press turns into a scroll', async ({
+    app,
+  }) => {
+    await press(app, TRACK_ROW, { driftY: 40 });
+    await app.waitForTimeout(HELD);
+
+    expect(await selectionCount(app)).toBeNull();
+    expect(await panel(app, 'track-list')).toBeNull();
+  });
+});
+
+test.describe('a hold anywhere else still opens the menu', () => {
+  test.beforeEach(async ({ app }) => {
+    await app.setViewportSize(PHONE);
+    await app.getByTestId('tab-albums').click();
+    await expect(app.getByTestId('main-content')).toHaveAttribute(
+      'data-active-view',
+      'albums',
+    );
+  });
+
+  test.afterEach(async ({ app }) => {
+    await app.setViewportSize({ width: 1440, height: 900 });
+  });
+
+  test('reaches the delegated handler and opens the real menu', async ({
+    app,
+  }) => {
+    // The property that let #63 reassign the hold without touching one
+    // of the fourteen context menus: unclaimed, it is what it was.
+    // Without this half, breaking all of them passes the suite.
+    await expect.poll(() => panel(app, 'cover-grid')).toBeNull();
+
+    await press(app, { host: 'cover-grid', inner: '[role="option"]' });
+
+    await expect
+      .poll(() => panel(app, 'cover-grid'), { timeout: HELD + 2000 })
+      .toMatchObject({ role: 'menu' });
+
+    // The same panel Shift+F10 opens, items and all -- not an empty
+    // popup that happened to become visible.
+    expect((await panel(app, 'cover-grid'))?.items).toBeGreaterThan(0);
+  });
+});

frontend/index.ts

@@ -70,7 +70,7 @@ import '@store/theme-store';
 // registers the document keydown listener for global shortcuts.
 import './src/services/keyboard-shortcut-service';
 import { activateView, deactivateView } from '@utils/view-lifecycle';
-import { installLongPressContextMenu } from '@utils/long-press';
+import { installTouchGestures } from '@utils/touch-gestures';
 import { openQueue, queuePanelElement } from '@utils/open-queue';
 import { installTopBarFit } from './src/services/top-bar-fit';
 import {
@@ -87,10 +87,13 @@ setBasePath('/dist/webawesome');
 // the session.
 registerBundledIcons();
 
-// The touch equivalent of a right-click, installed once for every menu
-// in the app rather than per component. Harmless on a desktop: it acts
-// on `pointerType === 'touch'` only.
-installLongPressContextMenu();
+// Every touch gesture in the app, installed once rather than per
+// component (plan 019). Harmless on a desktop: it acts on
+// `pointerType === 'touch'` only, per event, so a mouse on a
+// touchscreen keeps click-selects / double-click-plays on the very
+// same row. An unclaimed long press still becomes a `contextmenu`,
+// which is what leaves all fourteen menus untouched by #63.
+installTouchGestures();
 
 // The top bar decides what it can afford to show (#143). Here rather
 // than in a component because the bar is light DOM in index.html and

frontend/src/components/artist-details/artist-details.ts

@@ -14,6 +14,7 @@ import {
 import '@awesome.me/webawesome/dist/components/icon/icon.js';
 import '@components/cover-grid/cover-grid.js';
 import { designTokens } from '../../styles/tokens.css';
+import { backButton } from '../../styles/back-button.css';
 
 @customElement('artist-details')
 export class ArtistDetails extends LitElement {
@@ -40,7 +41,7 @@ export class ArtistDetails extends LitElement {
     /** Tracks the store's cached array reference to detect refreshes. */
     private lastAlbumsRef: library.Album[] | null = null;
 
-    static override styles = [designTokens, css`
+    static override styles = [designTokens, backButton, css`
         :host {
             display: flex;
             flex-direction: column;
@@ -66,31 +67,6 @@ export class ArtistDetails extends LitElement {
                 );
         }
 
-        .back-button {
-            display: flex;
-            align-items: center;
-            justify-content: center;
-            width: 32px;
-            height: 32px;
-            border: none;
-            border-radius: 50%;
-            background: var(
-                --yj-bg-overlay,
-                rgba(255, 255, 255, 0.06)
-            );
-            color: var(--yj-text-primary, #fff);
-            cursor: pointer;
-            flex-shrink: 0;
-            transition: background-color 0.15s ease;
-        }
-
-        .back-button:hover {
-            background: var(
-                --yj-bg-hover,
-                rgba(255, 255, 255, 0.12)
-            );
-        }
-
         .back-button wa-icon {
             font-size: 16px; /* back button — outside type scale */
         }

frontend/src/components/audio-player/seekbar/seek-bar.ts

@@ -45,18 +45,6 @@ export class SeekBar extends LitElement {
   private showRemaining: boolean = true;
 
   static override styles = [designTokens, waSliderLabel, css`
-    /* 12px below the phone breakpoint. The bottom bar's seek bar is
-       display:none there (016 B2 phase 1), so the only instance a
-       viewport media query can reach at that width is the full-screen
-       now-playing view's -- which is exactly the one a thumb uses.
-       The track size lives on wa-slider inside this shadow root, so a
-       custom property set by the host would not reach it. */
-    @media (max-width: 599px) {
-      wa-slider {
-        --track-size: 12px;
-      }
-    }
-
     wa-slider {
       --track-size: 6px;
       flex: 1;
@@ -80,6 +68,57 @@ export class SeekBar extends LitElement {
       background: var(--yj-bg-base, black);
     }
 
+    /* The phone's seek bar, and this block is last on purpose.
+
+       A media query adds no specificity, so this lived above the plain
+       "wa-slider" rule and lost to it at every width: the 12px track it
+       asks for had never once applied, and the bar measured 261x6 on
+       the device while the source said 12. That is index.css's rule
+       ("the phone section is last on purpose") met inside a component's
+       own stylesheet, and nothing renders differently in any tier here
+       to say so.
+
+       The bottom bar's seek bar is display:none below this width (016
+       B2 phase 1), so the only instance a viewport media query can
+       reach is the full-screen now-playing view's -- which is exactly
+       the one a thumb uses. The desktop bar keeps its 6px, where a
+       mouse is precise and the thickness is right.
+
+       The painted track and the thing you can hit are allowed to
+       differ, and a slider is the clearest case where they should: 12px
+       is a progress bar you can see, and 44px is the app's touch floor
+       (#56). A 44px-*thick* bar would be wrong-looking and would cost
+       the album art the vertical space #51 spent an issue recovering.
+
+       Two things about how the target is built.
+
+       The padding goes on ::part(slider) rather than on the host,
+       because that inner div is what carries the gesture -- it has the
+       listener and the touch-action: none, and it is exactly the host's
+       size, so padding the host would grow a box that does not take the
+       press.
+
+       The padding is asymmetric and the margins cancel it, so the row
+       does not grow by the difference. Both halves are measured: the
+       seek row is 19px (its clocks, not the track, decide that) and the
+       play button's top edge is 8px below it, so the target takes the
+       space *above*, where .art is a non-interactive div. Growing the
+       row instead cost the art 25px of 143. Verified on the device at
+       424x439: hit area 44px, painted track 12px, row still 19px, art
+       still 143px, 8px of clearance left under the play button, a press
+       26px above the track seeks, and a hit test on the play button's
+       top edge still reaches the play button. */
+    @media (max-width: 599px) {
+      wa-slider {
+        --track-size: 12px;
+      }
+
+      wa-slider::part(slider) {
+        padding-block: 28px 4px;
+        margin-block: -28px -4px;
+      }
+    }
+
     #seek-bar-container {
       display: flex;
       justify-content: space-between;

frontend/src/components/autotag-view/autotag-view.ts

@@ -250,13 +250,20 @@ export class AutotagView extends ViewLifecycleMixin(LitElement) {
             }
 
             /* Collapsible-section toggle used in the Pending header —
-               transparent button that inherits the header's type. */
+               transparent button that inherits the header's type.
+
+               187x**15** before this (#186), which was the smallest
+               control measured anywhere in the app until the column
+               arrows were counted. It is transparent and full-width
+               already, so the floor costs it a height and nothing
+               else. */
             .section-toggle {
                 display: flex;
                 align-items: center;
                 gap: 0.35rem;
                 flex: 1;
                 min-width: 0;
+                min-block-size: 44px;
                 padding: 0;
                 background: transparent;
                 border: 0;
@@ -274,6 +281,8 @@ export class AutotagView extends ViewLifecycleMixin(LitElement) {
                 color: var(--yj-text-tertiary, #888);
             }
 
+            /* 32x18, and it has no background until hover -- so the
+               padding out to a square target is invisible (#186). */
             .folders-menu-trigger {
                 background: transparent;
                 border: 0;
@@ -281,6 +290,8 @@ export class AutotagView extends ViewLifecycleMixin(LitElement) {
                 font-size: 1.1rem;
                 line-height: 1;
                 padding: 0.1rem 0.4rem;
+                min-inline-size: 44px;
+                min-block-size: 44px;
                 border-radius: 3px;
                 cursor: pointer;
             }
@@ -293,7 +304,10 @@ export class AutotagView extends ViewLifecycleMixin(LitElement) {
             .folders-refresh-trigger {
                 display: flex;
                 align-items: center;
+                justify-content: center;
                 font-size: 0.95rem;
+                min-inline-size: 44px;
+                min-block-size: 44px;
             }
 
             .folders-refresh-trigger:disabled {

frontend/src/components/config-page/config-field.ts

@@ -85,6 +85,21 @@ export class ConfigField extends LitElement {
             gap: 0.5em;
         }
 
+        /* Every control here meets the app's 44px touch floor (#186).
+
+           This is the shape every row in Settings uses, so it is the
+           one rule that covers the most controls -- and it is the
+           *cheapest* place to reach the floor, because there is no
+           overflow fit on this page. The page header's had one (#69),
+           which is why that pass had to grow padding and hand the
+           width back with a negative margin; here the control is a
+           block in a column and a taller box costs nothing but the
+           height it takes.
+
+           Measured on the reference device before this: the select
+           335x30, the text and number inputs the same, the browse
+           button 30 tall, the colour swatch 33x33 and the toggle
+           **34x19**. */
         input[type='text'],
         input[type='number'] {
             background: var(--yj-bg-elevated, #343a40);
@@ -95,6 +110,7 @@ export class ConfigField extends LitElement {
             font-size: 0.85em;
             font-family: inherit;
             min-width: 0;
+            min-block-size: 44px;
             flex: 1;
         }
 
@@ -117,6 +133,7 @@ export class ConfigField extends LitElement {
             font-size: 0.85em;
             font-family: inherit;
             cursor: pointer;
+            min-block-size: 44px;
             flex: 1;
         }
 
@@ -139,6 +156,7 @@ export class ConfigField extends LitElement {
             font-size: 0.85em;
             cursor: pointer;
             white-space: nowrap;
+            min-block-size: 44px;
         }
 
         button:hover {
@@ -158,8 +176,12 @@ export class ConfigField extends LitElement {
         }
 
         input[type='color'] {
-            width: 2.5em;
-            height: 2.5em;
+            /* border-box, or the 2px border makes this 48 and the
+               assertion below reads as passing by four pixels of
+               border rather than by the rule. */
+            box-sizing: border-box;
+            width: 44px;
+            height: 44px;
             border: 2px solid var(--yj-border, #444);
             border-radius: 4px;
             padding: 0;
@@ -187,12 +209,32 @@ export class ConfigField extends LitElement {
             display: flex;
             align-items: center;
             justify-content: space-between;
+            min-block-size: 44px;
         }
 
+        /* The toggle is the one control here whose target and paint
+           must differ, and it is also the one no sweep can see.
+
+           Its <input> is opacity: 0; width: 0; height: 0, so a
+           walk of every input on the page skips it as a zero-sized
+           node -- the thing a finger actually hits is this <label>,
+           which measured **34x19**. That is smaller than anything in
+           #186's original table and it is absent from it for exactly
+           that reason.
+
+           A 44px pill is not what a switch should look like, so the
+           box is 44px and the paint is not: .toggle-slider is a
+           2.5em x 1.4em child centred in it rather than an absolute
+           fill. The negative inline margins hand the extra width back
+           to the layout, so the pill stays flush with the right edge
+           of the inputs in the rows above it -- the header pass's
+           shape, used here for alignment rather than for a fit. */
         .toggle-switch {
-            position: relative;
-            width: 2.5em;
-            height: 1.4em;
+            display: grid;
+            place-items: center;
+            inline-size: 44px;
+            block-size: 44px;
+            margin-inline: calc((2.5em - 44px) / 2);
         }
 
         .toggle-switch input {
@@ -202,9 +244,10 @@ export class ConfigField extends LitElement {
         }
 
         .toggle-slider {
-            position: absolute;
+            position: relative;
             cursor: pointer;
-            inset: 0;
+            inline-size: 2.5em;
+            block-size: 1.4em;
             background: var(--yj-bg-overlay, #495057);
             border-radius: 1em;
             transition: background 0.2s;

frontend/src/components/config-page/config-page.ts

@@ -176,6 +176,13 @@ export class ConfigPage extends ViewLifecycleMixin(LitElement) {
             cursor: pointer;
             transition: background-color 0.15s ease;
             white-space: nowrap;
+            /* The app's 44px touch floor (#56, #186), stated once for
+               all 41 buttons this page renders rather than per class.
+               Height is free here: Settings has no overflow fit, so
+               the header's "only width is contested" rule does not
+               bind, and the two classes that need more than a height
+               say so below. */
+            min-block-size: 44px;
         }
 
         button:disabled {
@@ -509,11 +516,24 @@ export class ConfigPage extends ViewLifecycleMixin(LitElement) {
             margin: 0;
         }
 
+        /* The two column lists are the densest thing in the app, and
+           the density argument is why they are shaped the way they
+           are rather than simply grown (#186).
+
+           Measured on the reference device: the row was already
+           335x36 -- it is the controls *inside* it that were 16x16 and
+           **16x14**, the smallest anywhere in this app, 36 of them.
+           So the fix grows the controls into the row they already
+           occupy and only takes the row from 36 to 44, which over the
+           two lists (10 and 19 items) is 232px of extra scroll on a
+           439px screen. Growing each control to its own 44px row
+           instead would have cost four screens. */
         .column-item {
             display: flex;
-            align-items: center;
+            align-items: stretch;
             gap: 0.5em;
-            padding: 0.5em 0.75em;
+            padding: 0 0.75em;
+            min-block-size: 44px;
             border-bottom: 1px solid
                 var(--yj-border-subtle, #333);
             font-size: 0.85em;
@@ -531,8 +551,19 @@ export class ConfigPage extends ViewLifecycleMixin(LitElement) {
             color: var(--yj-text-tertiary, #888);
         }
 
+        /* A native checkbox cannot grow its hit area without growing
+           its paint, and a 44px checkbox is not what anyone wants. So
+           the target is the label instead: .column-label is a real
+           <label for> now, which makes the column's *name* the thing
+           you tap -- ~250x44 rather than 16x16.
+
+           That is the argument config-field already makes one file
+           over for its own labels: "a real label association also
+           makes the label text a click target for the control, which
+           is behaviour, not annotation". Here it is the whole fix. */
         .column-toggle {
             cursor: pointer;
+            align-self: center;
             accent-color: var(
                 --yj-accent,
                 #ffd43b
@@ -541,20 +572,32 @@ export class ConfigPage extends ViewLifecycleMixin(LitElement) {
 
         .column-label {
             flex: 1;
+            display: flex;
+            align-items: center;
+            cursor: pointer;
+            min-block-size: 44px;
         }
 
         .view-note {
             color: var(--yj-text-tertiary, #888);
             font-size: var(--yj-font-size-sm, 0.85rem);
             margin-left: auto;
+            /* The row stretches its children so the label can be a
+               full-height target; this is text, not a target. */
+            align-self: center;
         }
 
         .column-arrows {
             display: flex;
+            align-items: stretch;
             gap: 0.15em;
             margin-left: auto;
         }
 
+        /* 16x14 before this, and they carry background: none and a
+           transparent border -- so padding out to 44px grows the
+           target and changes nothing anyone can see until hover,
+           which is precisely what #186's Direction asks for. */
         .column-arrow-btn {
             background: none;
             border: 1px solid transparent;
@@ -564,6 +607,8 @@ export class ConfigPage extends ViewLifecycleMixin(LitElement) {
             font-size: 0.65em;
             line-height: 1;
             padding: 0.2em 0.35em;
+            min-inline-size: 44px;
+            min-block-size: 44px;
             transition:
                 color 0.15s,
                 border-color 0.15s;
@@ -711,6 +756,11 @@ export class ConfigPage extends ViewLifecycleMixin(LitElement) {
             padding: 0.2em 0.4em;
             letter-spacing: 2px;
             border-radius: 4px;
+            /* Square, so it needs the width too -- the shared rule
+               above only gives it a height. It was 31x31, and it is
+               the only route to "Remove library", which is the case
+               #55 settled one component over: the way out is 44px. */
+            min-inline-size: 44px;
         }
 
         .overflow-btn:hover {
@@ -2003,6 +2053,7 @@ export class ConfigPage extends ViewLifecycleMixin(LitElement) {
                                 class="column-item ${checked ? 'enabled' : 'disabled'}"
                             >
                                 <input
+                                    id="view-${v.id}"
                                     type="checkbox"
                                     class="column-toggle"
                                     aria-label="Show ${v.label} in the navigation"
@@ -2014,9 +2065,9 @@ export class ConfigPage extends ViewLifecycleMixin(LitElement) {
                         (e.target as HTMLInputElement).checked,
                     )}
                                 />
-                                <span class="column-label">
+                                <label class="column-label" for="view-${v.id}">
                                     ${v.label}
-                                </span>
+                                </label>
                                 ${note
                     ? html`<span class="view-note">${note}</span>`
                     : nothing}
@@ -2212,6 +2263,7 @@ export class ConfigPage extends ViewLifecycleMixin(LitElement) {
                                 class="column-item ${checked ? 'enabled' : 'disabled'}"
                             >
                                 <input
+                                    id="column-${id}"
                                     type="checkbox"
                                     class="column-toggle"
                                     aria-label="Show the ${columnLabel} column"
@@ -2222,11 +2274,12 @@ export class ConfigPage extends ViewLifecycleMixin(LitElement) {
                                             id,
                                         )}
                                 />
-                                <span
+                                <label
                                     class="column-label"
+                                    for="column-${id}"
                                 >
                                     ${columnLabel}
-                                </span>
+                                </label>
                                 <span
                                     class="column-arrows"
                                 >

frontend/src/components/config-page/download-clients.ts

@@ -8,6 +8,7 @@ import '@awesome.me/webawesome/dist/components/switch/switch.js';
 import '@awesome.me/webawesome/dist/components/spinner/spinner.js';
 import '@awesome.me/webawesome/dist/components/callout/callout.js';
 import { designTokens } from '../../styles/tokens.css';
+import { waTouchFloor } from '../../styles/wa-touch-floor.css';
 import type {
     DownloadDescriptor,
     DownloadProvider,
@@ -136,6 +137,7 @@ export class DownloadClients extends LitElement {
 
     static override styles = [
         designTokens,
+        waTouchFloor,
         css`
             :host {
                 display: block;
@@ -239,12 +241,17 @@ export class DownloadClients extends LitElement {
                 margin-top: 0.4em;
             }
 
+            /* The checkbox is 16x16 and cannot grow without becoming
+               a 44px checkbox, but it is already wrapped in the label
+               that names it -- so the label is the target and only
+               needs the height (#186). Eight of them. */
             .format-option {
                 display: flex;
                 align-items: center;
                 gap: 0.4em;
                 font-size: 0.9em;
                 cursor: pointer;
+                min-block-size: 44px;
             }
         `,
     ];

frontend/src/components/config-page/shortcut-capture.ts

@@ -25,6 +25,11 @@ export class ShortcutCapture extends LitElement {
         :host {
             display: inline-block;
         }
+        /* 80x25, twenty-six of them -- the most numerous control on
+           the Settings page after the column lists (#186). The floor
+           is a height here and nothing else: the width was already
+           past it, and the type stays where it is so a shortcut still
+           reads as a key rather than as a button. */
         button {
             font-family: inherit;
             font-size: var(--yj-text-sm, 13px);
@@ -35,6 +40,7 @@ export class ShortcutCapture extends LitElement {
             color: var(--yj-text-primary, #eee);
             cursor: pointer;
             min-width: 80px;
+            min-height: 44px;
             text-align: center;
             transition:
                 border-color 0.15s,
@@ -61,6 +67,11 @@ export class ShortcutCapture extends LitElement {
                 opacity: 0.7;
             }
         }
+        /* Reset renders only for a rebound shortcut, so a sweep of a
+           freshly-installed app never sees it -- it is not in #186's
+           tables for that reason, and it is a touch target the moment
+           anybody uses the feature. It also has no background, so the
+           padding out to 44px is invisible. */
         .reset-btn {
             font-size: var(--yj-text-xs, 11px);
             padding: 2px 6px;
@@ -69,7 +80,8 @@ export class ShortcutCapture extends LitElement {
             background: transparent;
             color: var(--yj-text-tertiary, #888);
             cursor: pointer;
-            min-width: auto;
+            min-width: 44px;
+            min-height: 44px;
             opacity: 0;
             transition: opacity 0.15s;
         }
@@ -79,6 +91,17 @@ export class ShortcutCapture extends LitElement {
         .reset-btn:hover {
             color: var(--yj-accent-text, #ffd43b);
         }
+        /*
+         * Reset is the only way to put a rebound shortcut back, so where
+         * the device has no hover it is always visible rather than an
+         * invisible button holding its hit area. The inverse of #68's
+         * rule, which applies where the hover control is redundant.
+         */
+        @media not all and (hover: hover) {
+            .reset-btn {
+                opacity: 1;
+            }
+        }
     `;
 
     private handleClick = () => {

frontend/src/components/downloads-view/downloads-view.ts

@@ -93,8 +93,26 @@ export class DownloadsView extends ViewLifecycleMixin(LitElement) {
                 border-bottom: 1px solid var(--yj-bg-overlay, rgba(255, 255, 255, 0.08));
             }
 
+            /* 85x34 and 96x34 before this (#186). A tab is the only
+               route to the panel it names, so it is the last control
+               that should be hard to hit -- and the underline that
+               marks the active one is drawn on the bottom border,
+               which a taller box moves further from the label. So the
+               height goes on *padding*, keeping the border against
+               the label rather than 10px below a centred one.
+
+               The min-size is the floor and is not redundant: padding
+               alone made this 44px here and **43px in CI**, because
+               the total is 13 + 13 + 2 + whatever line box the font
+               gives 13px text, and ubuntu:24.04's is a pixel shorter
+               than this machine's. A height computed from a font's
+               line box is not a height you control -- the same
+               mistake #195 made about a layout property measured on
+               one engine, one layer down, and caught here by the test
+               rather than by a person. */
             .tab {
-                padding: 8px 14px;
+                min-block-size: 44px;
+                padding: 13px 14px;
                 font-size: 13px;
                 font-weight: 600;
                 color: var(--yj-text-secondary, #b3b3b3);

frontend/src/components/explore-album-details/explore-album-details.ts

@@ -2,6 +2,7 @@ import { LitElement, html, css, nothing } from 'lit';
 import { customElement, property, state, query } from 'lit/decorators.js';
 import { classMap } from 'lit/directives/class-map.js';
 import { designTokens } from '../../styles/tokens.css';
+import { backButton } from '../../styles/back-button.css';
 import { srOnly } from '../../styles/sr-only.css';
 import { unownedLabel, unownedStyles } from '@utils/ownership';
 import {
@@ -355,6 +356,7 @@ export class ExploreAlbumDetails extends LitElement implements ContextMenuHost {
 
     static override styles = [
         designTokens,
+        backButton,
         exploreLinkStyles,
         contextMenuStyles,
         srOnly,
@@ -379,25 +381,6 @@ export class ExploreAlbumDetails extends LitElement implements ContextMenuHost {
                     var(--yj-border-subtle, rgba(255, 255, 255, 0.06));
             }
 
-            .back-button {
-                display: flex;
-                align-items: center;
-                justify-content: center;
-                width: 32px;
-                height: 32px;
-                border: none;
-                border-radius: 50%;
-                background: var(--yj-bg-overlay, rgba(255, 255, 255, 0.06));
-                color: var(--yj-text-primary, #fff);
-                cursor: pointer;
-                flex-shrink: 0;
-                transition: background-color 0.15s ease;
-            }
-
-            .back-button:hover {
-                background: var(--yj-bg-hover, rgba(255, 255, 255, 0.12));
-            }
-
             .back-button wa-icon {
                 font-size: 16px;
             }

frontend/src/components/explore-artist-details/explore-artist-details.ts

@@ -3,6 +3,7 @@ import { LitElement, html, css, nothing } from 'lit';
 import { customElement, property, state, query } from 'lit/decorators.js';
 import { classMap } from 'lit/directives/class-map.js';
 import { designTokens } from '../../styles/tokens.css';
+import { backButton } from '../../styles/back-button.css';
 import {
     LookupArtist,
     BrowseReleaseGroups,
@@ -266,6 +267,7 @@ export class ExploreArtistDetails extends LitElement implements ContextMenuHost
 
     static override styles = [
         designTokens,
+        backButton,
         exploreLinkStyles,
         contextMenuStyles,
         unownedStyles,
@@ -289,25 +291,6 @@ export class ExploreArtistDetails extends LitElement implements ContextMenuHost
                     var(--yj-border-subtle, rgba(255, 255, 255, 0.06));
             }
 
-            .back-button {
-                display: flex;
-                align-items: center;
-                justify-content: center;
-                width: 32px;
-                height: 32px;
-                border: none;
-                border-radius: 50%;
-                background: var(--yj-bg-overlay, rgba(255, 255, 255, 0.06));
-                color: var(--yj-text-primary, #fff);
-                cursor: pointer;
-                flex-shrink: 0;
-                transition: background-color 0.15s ease;
-            }
-
-            .back-button:hover {
-                background: var(--yj-bg-hover, rgba(255, 255, 255, 0.12));
-            }
-
             .back-button wa-icon {
                 font-size: 16px;
             }

frontend/src/components/explore-view/explore-view.ts

@@ -256,15 +256,18 @@ export class ExploreView extends ViewLifecycleMixin(LitElement) implements Conte
                 margin-bottom: 10px;
             }
 
+            /* 89x26 and 79x26 before this (#186). */
             .search-mode-tab {
                 display: inline-flex;
                 align-items: center;
+                justify-content: center;
                 gap: 6px;
                 background: none;
                 border: 1px solid transparent;
                 border-radius: 6px;
                 color: var(--yj-text-tertiary, #888);
                 cursor: pointer;
+                min-block-size: 44px;
                 padding: 5px 12px;
                 font-size: var(--yj-text-sm);
                 font-family: inherit;
@@ -289,7 +292,7 @@ export class ExploreView extends ViewLifecycleMixin(LitElement) implements Conte
                 border-radius: 6px;
                 padding: 0 12px;
                 gap: 8px;
-                height: 36px;
+                min-height: 44px;
                 max-width: 520px;
                 transition: border-color 0.15s ease;
             }
@@ -364,8 +367,15 @@ export class ExploreView extends ViewLifecycleMixin(LitElement) implements Conte
                 flex-shrink: 0;
             }
 
+            /* The input measured 325x**18** and the box around it 36,
+               which is two faults rather than one (#186): the row was
+               under the floor, and the input did not fill it, so eight
+               of those pixels were not a target at all. The container
+               is 44 and the input stretches to it -- a tap anywhere in
+               the box now lands on the input rather than beside it. */
             input {
                 flex: 1;
+                align-self: stretch;
                 background: none;
                 border: none;
                 outline: none;
@@ -379,15 +389,21 @@ export class ExploreView extends ViewLifecycleMixin(LitElement) implements Conte
                 color: var(--yj-text-tertiary, #888);
             }
 
+            /* No background until hover, so the target grows and the
+               glyph does not. It is inside a 44px box already, hence
+               the width alone. */
             .clear-button {
                 display: flex;
                 align-items: center;
                 justify-content: center;
+                align-self: stretch;
                 background: none;
                 border: none;
                 color: var(--yj-text-tertiary, #888);
                 cursor: pointer;
                 padding: 0;
+                min-inline-size: 44px;
+                margin-inline-end: -12px;
                 font-size: var(--yj-text-sm);
                 flex-shrink: 0;
             }

frontend/src/components/genre-details/genre-details.ts

@@ -15,6 +15,7 @@ import { describeError } from '@utils/describe-error';
 import '@awesome.me/webawesome/dist/components/icon/icon.js';
 import '@components/track-list/track-list.js';
 import { designTokens } from '../../styles/tokens.css';
+import { backButton } from '../../styles/back-button.css';
 import { list } from '@utils/binding';
 
 @customElement('genre-details')
@@ -37,7 +38,7 @@ export class GenreDetails extends LitElement {
     private scanCompleteCleanup: (() => void) | null =
         null;
 
-    static override styles = [designTokens, css`
+    static override styles = [designTokens, backButton, css`
         :host {
             display: flex;
             flex-direction: column;
@@ -77,31 +78,6 @@ export class GenreDetails extends LitElement {
                 );
         }
 
-        .back-button {
-            display: flex;
-            align-items: center;
-            justify-content: center;
-            width: 32px;
-            height: 32px;
-            border: none;
-            border-radius: 50%;
-            background: var(
-                --yj-bg-overlay,
-                rgba(255, 255, 255, 0.06)
-            );
-            color: var(--yj-text-primary, #fff);
-            cursor: pointer;
-            flex-shrink: 0;
-            transition: background-color 0.15s ease;
-        }
-
-        .back-button:hover {
-            background: var(
-                --yj-bg-hover,
-                rgba(255, 255, 255, 0.12)
-            );
-        }
-
         .back-button wa-icon {
             font-size: 16px; /* back button — outside type scale */
         }

frontend/src/components/home-view/home-view.ts

@@ -166,7 +166,7 @@ export class HomeView extends ViewLifecycleMixin(LitElement) {
              * gated on the device having hover rather than on width.  A
              * touch long-press synthesises a hover state in the WebView,
              * so on a phone it flashed into view during the 500ms hold
-             * that utils/long-press.ts is measuring for a context menu —
+             * that utils/touch-gestures.ts is measuring for a long press —
              * a control appearing because you were reaching for a
              * different one.  A phone user taps the album and plays from
              * the detail view, so there is nothing to replace it with.

frontend/src/components/library-filter/library-filter.ts

@@ -23,8 +23,14 @@ export class LibraryFilter extends LitElement {
             align-items: center;
         }
 
+        /* 120x32 on the reference device (#186). This control has two
+           placements since #57 -- the desktop top bar and Settings ->
+           Libraries -- and it is the only route to setSelectedLibrary
+           in either, so it is one of the controls #148 argued must not
+           simply be taken away. It is one component, so it reaches the
+           floor in one place. */
         select {
-            height: 32px;
+            min-height: 44px;
             padding: 0 8px;
             border-radius: 6px;
             border: 1px solid

frontend/src/components/now-playing-view/now-playing-view.ts

@@ -105,6 +105,19 @@ export class NowPlayingView extends LitElement {
             color: var(--yj-text-secondary, #adb5bd);
         }
 
+        /* The art and the names are one block, so that a short
+           screen can lay them out side by side without either of them
+           knowing about the other's box. Vertically it is exactly what
+           the host used to do -- same gap, art flexible, names fixed --
+           so the tall layout is unchanged. */
+        .stack {
+            display: flex;
+            flex-direction: column;
+            gap: 0.75em;
+            flex: 1 1 auto;
+            min-height: 0;
+        }
+
         .art {
             flex: 1 1 auto;
             display: flex;
@@ -113,38 +126,89 @@ export class NowPlayingView extends LitElement {
             min-height: 0;
         }
 
-        .art img,
-        .art .placeholder {
-            /* Square, and never taller than the room left over: the
-               art is the one thing here that would happily push the
-               transport off the bottom of a short phone.
+        .art img {
+            /* Square, and never larger than the room left over --
+               where "square" is a property of what is painted and not
+               just of what was asked for.
 
-               **max-height is what actually keeps that promise**, and
-               it was missing. With a definite width and
-               a 1:1 aspect-ratio the height is *derived from the width*
-               and is bounded by nothing: at the reference device's
-               424x439 that is a 263px square (60vh) in a box with far
-               less than 263px left, so the art overflowed its own
-               centred flex item and drew over the header above and the
-               title below it. The comment claimed this was handled;
-               60vh is a bound on the *viewport*, not on the room left
-               over, and those differ by however much chrome is above
-               and below.
+               The previous rule asked for a square and did not get
+               one. width: min(100%, 60vh) makes the width definite,
+               aspect-ratio: 1 derives the height from it, and
+               max-height: 100% then clamps that height **without
+               re-deriving the width** -- which is how the
+               aspect-ratio property is specified to behave, unlike
+               an intrinsic ratio. So whenever the room left over was
+               shorter than the box was wide, the art was drawn as a
+               letterbox strip and object-fit: cover cropped the
+               cover to it. Measured on the reference device at
+               424x439: **264x53**, a 5:1 band of a square image.
 
-               Pre-existing -- screenshotted on main -- and made acute
-               by #56, which gives the transport 95px more than it had.
-               Found by reading a screenshot, which is the only tier
-               that can see it: nothing fails, nothing overflows the
-               *shell*, and every control is still hittable. */
-            width: min(100%, 60vh);
+               That is not only the phone. The leftover exceeds the
+               width only above ~843px of viewport, so every height
+               from ~500 to ~843 -- most phones, and any small window
+               -- drew a cropped strip too.
+
+               Both maxes with auto sizes is the fix, and it is the
+               replaced-element path rather than the aspect-ratio
+               one: the used size preserves the ratio under *both*
+               bounds (CSS2.1 10.4), so the art is square at every
+               height. Checked against Chrome 113 itself -- the
+               device's engine -- at column heights of 288, 300, 451,
+               600 and 800: square at all five, where the old rule
+               cropped at four.
+
+               A corollary worth knowing: auto will not upscale past
+               the image's natural size, and the largest tier
+               saveCoverArt keeps is 400px. Drawing it larger was
+               upscaling, so nothing is lost. */
+            max-width: 100%;
             max-height: 100%;
+            width: auto;
+            height: auto;
             aspect-ratio: 1;
             object-fit: cover;
             border-radius: 12px;
             background-color: var(--yj-bg-elevated, #343a40);
         }
 
+        /* The placeholder is not a replaced element, so it cannot use
+           the rule above: with no intrinsic size, auto/auto collapses
+           it to its icon -- measured at 13x58 in Chrome 113, which is
+           neither square nor the art's size.
+
+           So it is sized from the height, and then bounded by the
+           width in the one way a box like this can be. A non-replaced
+           element cannot express "the largest square that fits" in a
+           single rule: aspect-ratio derives the second axis from the
+           first, and whichever max clamps it does not re-derive the
+           other, which is the same trap the image rule above is about.
+           Driving it from the height alone is right until the column
+           is taller than it is wide -- ~843px of viewport, which is a
+           tall phone and #51's other named device -- and there it went
+           380x484.
+
+           max-height in viewport units is what closes it, and it is
+           sound here for the reason 60vh was not: this view is a
+           phone-width detail view, so its content box really is the
+           viewport less the host's own 1rem gutters. It is a *max*, so
+           the failure mode if that ever stopped being true is a square
+           bounded slightly early rather than a crop. rem and not em --
+           this box sets font-size: 3rem for the icon, so 2em here
+           would be 96px. */
         .art .placeholder {
+            height: 100%;
+            width: auto;
+            max-width: 100%;
+            max-height: calc(100vw - 2rem);
+            /* A flex item's automatic minimum is its content, so
+               without this the icon's own width becomes a floor and
+               the box goes wider than it is tall the moment the row is
+               shorter than the icon -- which is exactly the state a
+               job band puts this screen in. */
+            min-width: 0;
+            aspect-ratio: 1;
+            border-radius: 12px;
+            background-color: var(--yj-bg-elevated, #343a40);
             display: flex;
             align-items: center;
             justify-content: center;
@@ -232,6 +296,60 @@ export class NowPlayingView extends LitElement {
             color: var(--yj-text-secondary, #adb5bd);
             text-align: center;
         }
+
+        /* Below 500px of viewport the art and the names sit side by
+           side, and that is the whole of this screen's answer to a
+           short phone (#51).
+
+           The stacked layout cannot be rescued by sizing alone. Its
+           budget is fixed -- 48px of header, 143px of transport since
+           #64, 78px of names, 68px of padding and gaps -- so the art
+           gets height - 386, which on the reference device's 424x439
+           is **53px**. #172 measured 39px before #64 and named the
+           two options: give the art a floor and let the block scroll,
+           or reflow. A floor scrolls the transport off the bottom,
+           and "controls never scroll off" is #51's own Direction and
+           plan 018's promise -- so it is the reflow.
+
+           Sideways the art is bounded by the row's height rather than
+           by the column's leftover, which is the whole gain: the same
+           439px screen goes from a 53px sliver to **143px**, measured
+           on the device, with nothing scrolling and the transport
+           untouched.
+
+           500 is where the two layouts cross rather than a round
+           number. In a row the art is height - 296 and the names get
+           what is left of 392px, so the names hold 176px at exactly
+           500 and less above it; stacked, the art is height - 386,
+           which passes 176px at 562. Below 500 the row is the bigger
+           art *and* the readable one -- above it the column is, which
+           is why a tall phone (a Pixel 7's ~869) keeps the layout it
+           has. Unverified on that device: none was attached.
+
+           It is keyed on height alone, not on the phone's width,
+           because it is an answer to vertical room -- a 900x450 window
+           has the same problem and the same fix. */
+        @media (max-height: 500px) {
+            .stack {
+                flex-direction: row;
+                align-items: center;
+            }
+
+            /* A square of the row's height. The box has to carry the
+               ratio here rather than the image, because in a row the
+               art's width is what the ratio has to produce -- and the
+               image's own rule then fits it to a box that is already
+               square. */
+            .art {
+                flex: 0 1 auto;
+                height: 100%;
+                aspect-ratio: 1;
+            }
+
+            .meta {
+                flex: 1 1 auto;
+            }
+        }
     `];
 
     private back() {
@@ -283,55 +401,57 @@ export class NowPlayingView extends LitElement {
         return html`
             ${this.renderHeader()}
 
-            <div class="art">
-                ${art
-                    ? html`<img
-                          src=${art}
-                          alt=""
-                          decoding="async"
-                          data-testid="npv-art"
-                      />`
-                    : html`<div class="placeholder" aria-hidden="true">
-                          <wa-icon name="compact-disc"></wa-icon>
-                      </div>`}
-            </div>
-
-            <div class="meta">
-                <div class="names">
-                    <h2 class="title" data-testid="npv-title">
-                        ${track.title || track.fileName}
-                    </h2>
-                    <p class="artist">
-                        ${creditLink(
-                            creditStore.credits(track.recordingMbid),
-                            track.artist,
-                            track.artistMbid,
-                        )}
-                    </p>
-                    ${track.album
-                        ? html`<p class="album">
-                              ${albumLink(
-                                  track.album,
-                                  track.releaseGroupMbid,
-                                  undefined,
-                                  track.artist,
-                              )}
-                          </p>`
-                        : nothing}
+            <div class="stack">
+                <div class="art">
+                    ${art
+                        ? html`<img
+                              src=${art}
+                              alt=""
+                              decoding="async"
+                              data-testid="npv-art"
+                          />`
+                        : html`<div class="placeholder" aria-hidden="true">
+                              <wa-icon name="compact-disc"></wa-icon>
+                          </div>`}
                 </div>
 
-                <button
-                    type="button"
-                    class="favorite ${favorited ? 'on' : ''}"
-                    data-testid="npv-favorite"
-                    aria-pressed=${favorited ? 'true' : 'false'}
-                    aria-label=${favorited
-                        ? `Remove ${track.title} from ${this.favCtrl.playlistName}`
-                        : `Add ${track.title} to ${this.favCtrl.playlistName}`}
-                    @click=${this.toggleFavorite}
-                >
-                    <wa-icon name=${this.favCtrl.iconFor(favorited)}></wa-icon>
-                </button>
+                <div class="meta">
+                    <div class="names">
+                        <h2 class="title" data-testid="npv-title">
+                            ${track.title || track.fileName}
+                        </h2>
+                        <p class="artist">
+                            ${creditLink(
+                                creditStore.credits(track.recordingMbid),
+                                track.artist,
+                                track.artistMbid,
+                            )}
+                        </p>
+                        ${track.album
+                            ? html`<p class="album">
+                                  ${albumLink(
+                                      track.album,
+                                      track.releaseGroupMbid,
+                                      undefined,
+                                      track.artist,
+                                  )}
+                              </p>`
+                            : nothing}
+                    </div>
+
+                    <button
+                        type="button"
+                        class="favorite ${favorited ? 'on' : ''}"
+                        data-testid="npv-favorite"
+                        aria-pressed=${favorited ? 'true' : 'false'}
+                        aria-label=${favorited
+                            ? `Remove ${track.title} from ${this.favCtrl.playlistName}`
+                            : `Add ${track.title} to ${this.favCtrl.playlistName}`}
+                        @click=${this.toggleFavorite}
+                    >
+                        <wa-icon name=${this.favCtrl.iconFor(favorited)}></wa-icon>
+                    </button>
+                </div>
             </div>
 
             <div class="transport">

frontend/src/components/page-header/page-header.ts

@@ -298,6 +298,47 @@ export class PageHeader extends LitElement {
                 flex-shrink: 0;
             }
 
+            /* Every control in this header meets the app's 44px touch
+               floor -- the number #56 set for the transport and the
+               queue header already keeps (#186).
+
+               It is min-size rather than padding with a negative
+               margin, which is what the seek bar needed (#187), and
+               the difference is worth stating because it decides
+               whether targets can collide. There the painted track had
+               to stay thin, so the target was grown past its own box
+               and had to be checked against its neighbours. Here the
+               control *is* the target: the boxes are flex items, so
+               the gap keeps them apart and no two can overlap by
+               construction.
+
+               There is no phone branch. With the target being the box,
+               a 44px control on a desktop is merely large, and a
+               second declaration of what a phone shows is a second
+               thing to keep in step -- which is the reason this
+               component has never had one. It also avoids a media
+               query that no tier here renders, which is exactly how
+               the seek bar's phone rule came to be dead for months.
+
+               **The height is the box and the width is not**, and that
+               asymmetry is the whole of what the overflow fit below
+               cares about. That pass measures inline size, so a taller
+               control costs it nothing and a wider one costs it
+               directly. Growing the two square controls to 44px wide
+               added 22px, which fits at every width Chromium was
+               checked at and clipped the overflow trigger at 320px in
+               **WebKit** -- the engine closest to what actually ships,
+               and the one no machine here can run. So the horizontal
+               half is padding with the margin cancelling it, which is
+               what the issue asked for in the first place: the target
+               grows and the layout does not.
+
+               The cost is that a horizontal target can now overlap a
+               neighbour, which the box version could not. The arrow's
+               is deliberately lopsided for the seek bar's reason
+               (#187): the select is 6px to its left and there is open
+               space to its right, so it takes the side with nothing to
+               steal from. */
             .sort select {
                 font: inherit;
                 color: inherit;
@@ -306,6 +347,7 @@ export class PageHeader extends LitElement {
                 border-radius: 4px;
                 padding: 3px 6px;
                 cursor: pointer;
+                min-block-size: 44px;
             }
 
             .sort-dir {
@@ -318,6 +360,18 @@ export class PageHeader extends LitElement {
                 color: inherit;
                 cursor: pointer;
                 padding: 3px 5px;
+                /* 28x21 before this, the smallest control in the
+                   header and the only one that failed the floor in
+                   both directions.
+
+                   Vertically the box grows, because the header has the
+                   room and nothing measures it. Horizontally the box
+                   must not: 28 + 2 + 14 is a 44px target over a 28px
+                   layout box, weighted right because the select is 6px
+                   to the left. */
+                min-block-size: 44px;
+                padding-inline: 5px 21px;
+                margin-inline: 0 -16px;
             }
 
             .sort-dir:hover {
@@ -377,10 +431,20 @@ export class PageHeader extends LitElement {
                 gap: 6px;
                 white-space: nowrap;
                 flex-shrink: 0;
+                justify-content: center;
+                min-block-size: 44px;
             }
 
             .more-button {
                 padding: 6px 10px;
+                /* 38x27, and it is the route to every collapsed
+                   action, so it is the last control that should be
+                   hard to hit -- and the one WebKit clipped at 320px
+                   when this was 6px wider as a box. 38 + 3 + 3 is a
+                   44px target over a 38px layout box; the actions row
+                   has an 8px gap, so this one can be symmetric. */
+                padding-inline: 13px;
+                margin-inline: -3px;
             }
 
             /* The display: flex above outranks the UA stylesheet's

frontend/src/components/playlist-details/playlist-details.ts

@@ -71,6 +71,7 @@ import {
     exploreLinkStyles,
 } from '@utils/explore-link';
 import { designTokens } from '../../styles/tokens.css';
+import { backButton } from '../../styles/back-button.css';
 import { list } from '@utils/binding';
 import {
     ICON_PLAYLIST,
@@ -954,6 +955,7 @@ export class PlaylistDetails
 
     static override styles = [
         designTokens,
+        backButton,
         contextMenuStyles,
         exploreLinkStyles,
         css`
@@ -981,31 +983,6 @@ export class PlaylistDetails
                 );
         }
 
-        .back-button {
-            display: flex;
-            align-items: center;
-            justify-content: center;
-            width: 32px;
-            height: 32px;
-            border: none;
-            border-radius: 50%;
-            background: var(
-                --yj-bg-overlay,
-                rgba(255, 255, 255, 0.06)
-            );
-            color: var(--yj-text-primary, #fff);
-            cursor: pointer;
-            flex-shrink: 0;
-            transition: background-color 0.15s ease;
-        }
-
-        .back-button:hover {
-            background: var(
-                --yj-bg-hover,
-                rgba(255, 255, 255, 0.12)
-            );
-        }
-
         .back-button wa-icon {
             font-size: 16px;
         }

frontend/src/components/queue-panel/queue-panel.ts

@@ -13,6 +13,7 @@ import type { MenuSurface } from '../menu-surface/menu-surface';
 import '../menu-surface/menu-surface';
 import '@awesome.me/webawesome/dist/components/dropdown-item/dropdown-item.js';
 import { QueueController } from '@store/controllers/queue-controller';
+import { PHONE_QUERY } from '@utils/breakpoints';
 import { creditStore } from '@store/credit-store';
 import {
     describeQueueSource,
@@ -120,6 +121,25 @@ export class QueuePanel
     @property({ type: Boolean, reflect: true })
     overlay = false;
 
+    /**
+     * Phone width, from `matchMedia` rather than from a media query,
+     * because it decides whether the scrim *exists* (#171) —
+     * `job-band`'s rule, and a stylesheet cannot express it: a
+     * `display: none` scrim is still an element with a click handler.
+     *
+     * Below 600px the panel spans the whole content area, so the scrim
+     * has no uncovered pixels: measured at 424x439, host, panel and
+     * scrim are all 424x318 with the scrim entirely underneath. It dims
+     * nothing and dismisses nothing there, and the queue is a *screen*
+     * at that width anyway (#55) — back and a 44px close button are its
+     * ways out. Between 600 and 899 the panel is a 320px column of a
+     * wider content area, the scrim is reachable, and #24's
+     * tap-outside-to-close is real; that band is untouched.
+     */
+    @state() private phone = false;
+
+    private phoneQuery?: MediaQueryList;
+
     @state()
     private isDragging = false;
 
@@ -391,6 +411,8 @@ export class QueuePanel
             display: none;
         }
 
+        /* Overlay only, and above 600px only -- see the phone field,
+           which is where that half is decided (#171). */
         .scrim {
             position: absolute;
             inset: 0;
@@ -414,10 +436,10 @@ export class QueuePanel
 
             /* A screen's way out has to be hittable with a thumb.
                Measured at 424x439 before #55: these were **25x21px**,
-               and with the panel spanning the whole width the scrim
-               underneath has no uncovered pixels at all -- so it was
-               the only pointer route out of a full-screen surface.
-               Back answers it now as well, which is the other half.
+               and with the panel spanning the whole width there is no
+               scrim here at all (#171) -- so this is the only pointer
+               route out of a full-screen surface. Back answers it now
+               as well, which is the other half.
 
                Sized only in overlay mode: inline these sit in a 320px
                column beside the content, where a mouse is what reaches
@@ -639,23 +661,42 @@ export class QueuePanel
             text-overflow: ellipsis;
         }
 
+        /*
+         * The per-row remove is a hover affordance, and on a device
+         * without hover it is redundant rather than missing: the row's
+         * context menu is a bottom sheet since #60 and carries "Remove
+         * from Queue", so the action is one long-press away. An
+         * always-visible X would instead spend part of a 424px row on
+         * something already reachable. #68's treatment, for #68's reason.
+         *
+         * display:none outside the query rather than visibility:hidden:
+         * a hidden button still occupies its hit area and is still in
+         * the accessibility tree, so a phone would keep a target for a
+         * control it can never see.
+         */
         .remove-button {
-            background: none;
-            border: none;
-            color: var(--yj-text-tertiary, #888);
-            cursor: pointer;
-            padding: 4px;
-            display: flex;
-            align-items: center;
-            visibility: hidden;
+            display: none;
         }
 
-        .track-item:hover .remove-button {
-            visibility: visible;
-        }
+        @media (hover: hover) and (pointer: fine) {
+            .remove-button {
+                background: none;
+                border: none;
+                color: var(--yj-text-tertiary, #888);
+                cursor: pointer;
+                padding: 4px;
+                display: flex;
+                align-items: center;
+                visibility: hidden;
+            }
 
-        .remove-button:hover {
-            color: var(--yj-error-text, #ff8787);
+            .track-item:hover .remove-button {
+                visibility: visible;
+            }
+
+            .remove-button:hover {
+                color: var(--yj-error-text, #ff8787);
+            }
         }
 
         .list-area.drag-over {
@@ -828,6 +869,12 @@ export class QueuePanel
         // desktop width rather than the minimum.
         this.updateOverlayMode();
 
+        // Read here rather than in a field initialiser, so a test can
+        // install its own matchMedia before the element is created.
+        this.phoneQuery = window.matchMedia?.(PHONE_QUERY);
+        this.phone = this.phoneQuery?.matches ?? false;
+        this.phoneQuery?.addEventListener('change', this.onPhoneMedia);
+
         if (this.parentElement) {
             this.spaceObserver = new ResizeObserver(() =>
                 this.updateOverlayMode(),
@@ -870,6 +917,8 @@ export class QueuePanel
         this.creditsUnsub = undefined;
         this.spaceObserver?.disconnect();
         this.spaceObserver = undefined;
+        this.phoneQuery?.removeEventListener('change', this.onPhoneMedia);
+        this.phoneQuery = undefined;
         document.removeEventListener('keydown', this.onOverlayKeydown);
         document.removeEventListener(
             'mousemove',
@@ -936,6 +985,10 @@ export class QueuePanel
         this.overlay = available - this.panelWidth < MAIN_PANEL_FLOOR;
     };
 
+    private onPhoneMedia = (e: MediaQueryListEvent): void => {
+        this.phone = e.matches;
+    };
+
     /**
      * Escape closes a scrimmed overlay, which is the one keyboard rule
      * every dialog in this app already follows.
@@ -1972,7 +2025,7 @@ export class QueuePanel
         const tracks = this.queue.tracks;
 
         return html`
-            ${this.overlay
+            ${this.overlay && !this.phone
                 ? html`<div
                       class="scrim"
                       part="scrim"

frontend/src/components/search-dialog/search-trigger.ts

@@ -61,11 +61,32 @@ export class SearchTrigger extends LitElement {
                 display: inline-flex;
                 align-items: center;
                 justify-content: center;
-                /* The smallest a touch target should be. The header's
-                   own action buttons are smaller because they carry a
-                   label; this one is a glyph. */
-                min-width: 40px;
-                min-height: 40px;
+                /* The app's touch floor, from #56 -- and this is the
+                   control that should least have to argue for it: #57
+                   created it as the phone's replacement for the header
+                   search box, so it exists *only* where there is a
+                   thumb.
+
+                   It shipped at 40px under a comment calling that "the
+                   smallest a touch target should be", which was the
+                   floor being restated four pixels short rather than a
+                   second opinion about it (#186). The rest of that
+                   comment said the header's own action buttons are
+                   smaller because they carry a label; they are 44px
+                   now too, so that no longer distinguishes anything.
+
+                   The extra width is a target rather than a box, for
+                   page-header's reason: this button sits in that
+                   header, whose overflow fit (#69) measures inline
+                   size, and four pixels there is four pixels the
+                   trigger for every collapsed action does not get at
+                   320px. Height is free -- nothing measures it. */
+                min-width: 44px;
+                min-height: 44px;
+                /* Border-box, so the 44 above is the whole target and
+                   the margin is what hands the four extra pixels back
+                   to the row. */
+                margin-inline: -2px;
                 padding: 0;
                 background: none;
                 border: 1px solid var(--yj-border-subtle, #555);

frontend/src/components/selection-bar/selection-bar.ts

@@ -0,0 +1,176 @@
+import { LitElement, html, css, nothing } from 'lit';
+import { customElement, property } from 'lit/decorators.js';
+import '@awesome.me/webawesome/dist/components/icon/icon.js';
+import { designTokens } from '../../styles/tokens.css';
+import { ICON_MORE_ACTIONS } from '@utils/icon-language';
+
+/**
+ * What a selection can have done to it, while a finger is holding one.
+ *
+ * This is the context menu (plan 019, #63). Not a second surface
+ * beside it — the same actions, contextualised to whatever is
+ * selected, in the shape Android puts them in.
+ *
+ * #63 asked for a double tap to open the menu instead. That mapping
+ * costs the app's primary interaction 250ms on every play, because the
+ * first tap of a double tap is indistinguishable from a single tap
+ * until the interval expires, and playing a track is ~100ms end to end
+ * on the reference device. So there is no double tap: a long press
+ * selects, this says what can be done, and the sheet behind "More" is
+ * the same `menu-surface` every other menu in the app opens.
+ *
+ * Four things about it are load-bearing.
+ *
+ * **It is presentational.** It takes a count and a list of actions and
+ * emits `selection-action` / `selection-exit`; it holds no selection
+ * and calls no store. The host already owns a `SelectionController`
+ * and an action handler, and a bar that reached for either would be a
+ * second definition of what "play the selection" means — the fault
+ * `utils/library-status.ts` exists to have fixed one feature over.
+ *
+ * **The bar is only what fits, and "More" is the rest.** A context
+ * menu can be nine items because it is a sheet; a bar is one row on a
+ * 424px screen. So the host passes the two or three worth a thumb and
+ * the overflow opens the menu it already renders, which is what keeps
+ * every action reachable at every size — plan 018's promise, and the
+ * reason this cannot simply drop the long tail.
+ *
+ * **Its controls are 44px** (#56, #186), and the count is a live
+ * region: the number changes under the user's finger as they tap rows,
+ * and nothing else on screen announces it.
+ *
+ * **It renders nothing at zero.** The mode ends when the last row is
+ * deselected — `SelectionController.toggleInMode` is where that is
+ * decided — so a bar with a count of none is a state this should never
+ * be asked to draw, and drawing it anyway would hide the fact that it
+ * has been.
+ */
+export interface SelectionAction {
+    id: string;
+    label: string;
+    icon: string;
+    danger?: boolean;
+}
+
+@customElement('selection-bar')
+export class SelectionBar extends LitElement {
+    static override styles = [
+        designTokens,
+        css`
+            :host {
+                display: block;
+            }
+
+            .bar {
+                display: flex;
+                align-items: center;
+                gap: 0.25em;
+                padding: 0.25em 0.5em;
+                background: var(--yj-bg-elevated, #343a40);
+                border-top: 1px solid var(--yj-border-subtle, #333);
+            }
+
+            .count {
+                flex: 1;
+                min-width: 0;
+                font-size: var(--yj-text-md);
+                font-weight: 600;
+                color: var(--yj-text-primary, #fff);
+                white-space: nowrap;
+                overflow: hidden;
+                text-overflow: ellipsis;
+            }
+
+            /* 44px, from #56 and #186 -- this is a bar a thumb uses. */
+            button {
+                display: flex;
+                align-items: center;
+                justify-content: center;
+                min-inline-size: 44px;
+                min-block-size: 44px;
+                padding: 0 0.5em;
+                border: none;
+                border-radius: 4px;
+                background: none;
+                color: var(--yj-text-primary, #fff);
+                font-family: inherit;
+                font-size: var(--yj-text-md);
+                cursor: pointer;
+            }
+
+            button:hover {
+                background: var(--yj-bg-overlay, #495057);
+            }
+
+            button.danger {
+                color: var(--yj-error-text, #ff8787);
+            }
+        `,
+    ];
+
+    /** How many items are selected. Zero renders nothing. */
+    @property({ type: Number }) count = 0;
+
+    /** The actions worth a thumb. The rest live behind "More". */
+    @property({ attribute: false }) actions: SelectionAction[] = [];
+
+    private emit(name: string, detail?: unknown) {
+        this.dispatchEvent(
+            new CustomEvent(name, { detail, bubbles: true, composed: true }),
+        );
+    }
+
+    override render() {
+        if (this.count <= 0) return nothing;
+
+        const noun = this.count === 1 ? 'track' : 'tracks';
+
+        return html`
+            <div class="bar" role="toolbar" aria-label="Selection actions">
+                <button
+                    aria-label="Leave selection"
+                    @click=${() => this.emit('selection-exit')}
+                >
+                    <wa-icon name="xmark"></wa-icon>
+                </button>
+                <span class="count" role="status" aria-live="polite">
+                    ${this.count.toLocaleString()} ${noun} selected
+                </span>
+                ${this.actions.map(
+                    (action) => html`
+                        <button
+                            class=${action.danger ? 'danger' : ''}
+                            aria-label=${action.label}
+                            title=${action.label}
+                            @click=${() =>
+                                this.emit('selection-action', { id: action.id })}
+                        >
+                            <wa-icon name=${action.icon}></wa-icon>
+                        </button>
+                    `,
+                )}
+                <button
+                    aria-label="More actions"
+                    @click=${(e: MouseEvent) => {
+                        const box = (
+                            e.currentTarget as HTMLElement
+                        ).getBoundingClientRect();
+
+                        this.emit('selection-more', {
+                            x: box.left,
+                            y: box.top,
+                        });
+                    }}
+                >
+                    <wa-icon name=${ICON_MORE_ACTIONS}></wa-icon>
+                </button>
+            </div>
+        `;
+    }
+}
+
+declare global {
+    interface HTMLElementTagNameMap {
+        'selection-bar': SelectionBar;
+    }
+}

frontend/src/components/smart-playlist-details/smart-playlist-details.ts

@@ -62,6 +62,7 @@ import {
 } from '@utils/explore-link';
 import '@components/smart-playlist-editor/smart-playlist-editor.js';
 import { designTokens } from '../../styles/tokens.css';
+import { backButton } from '../../styles/back-button.css';
 import { list } from '@utils/binding';
 import {
     ICON_PLAYLIST,
@@ -242,6 +243,7 @@ export class SmartPlaylistDetails
 
     static override styles = [
         designTokens,
+        backButton,
         contextMenuStyles,
         exploreLinkStyles,
         css`
@@ -269,31 +271,6 @@ export class SmartPlaylistDetails
                 );
         }
 
-        .back-button {
-            display: flex;
-            align-items: center;
-            justify-content: center;
-            width: 32px;
-            height: 32px;
-            border: none;
-            border-radius: 50%;
-            background: var(
-                --yj-bg-overlay,
-                rgba(255, 255, 255, 0.06)
-            );
-            color: var(--yj-text-primary, #fff);
-            cursor: pointer;
-            flex-shrink: 0;
-            transition: background-color 0.15s ease;
-        }
-
-        .back-button:hover {
-            background: var(
-                --yj-bg-hover,
-                rgba(255, 255, 255, 0.12)
-            );
-        }
-
         .back-button wa-icon {
             font-size: 16px;
         }

frontend/src/components/track-details/track-details.ts

@@ -529,6 +529,44 @@ export class TrackDetails extends LitElement {
             background: var(--yj-error, #e03131);
         }
 
+        /*
+         * Both are the *only* route to changing or removing a track's
+         * cover art, so where the device has no hover they are always
+         * visible rather than hidden — the inverse of #68's rule, which
+         * applies where the hover control is redundant. Revealed by
+         * opacity, so what is on screen is what the desktop reveal shows
+         * and nothing about the layout moves.
+         */
+        @media not all and (hover: hover) {
+            /* The × is genuinely the only route to removing the art, so
+               on a device that cannot hover it is simply always there.
+
+               The pen is not: .cover-art-edit carries the click that
+               opens the file picker, so tapping the artwork already
+               worked while the overlay was invisible. It is a discovery
+               hint — and paying for discovery by covering the artwork
+               being edited in 50% black, permanently, on every touch
+               device, is heavier than the hint is worth. It becomes a
+               corner chip in the remove button's own visual language
+               instead: same size, same disc, same alpha. */
+            .cover-art-remove {
+                opacity: 1;
+            }
+
+            .cover-art-overlay {
+                opacity: 1;
+                inset: auto 4px 4px auto;
+                width: 24px;
+                height: 24px;
+                border-radius: 50%;
+                background: rgba(0, 0, 0, 0.7);
+            }
+
+            .cover-art-overlay wa-icon {
+                font-size: 14px;
+            }
+        }
+
         /* Error message */
         .error-message {
             flex: 1;

frontend/src/components/track-list/track-list.ts

@@ -10,6 +10,9 @@ import {
 } from 'lit/decorators.js';
 import { SelectionController } from '@utils/selection-controller';
 import type { SelectionHost } from '@utils/selection-controller';
+import type { GestureEvent } from '@utils/touch-gestures';
+import '@components/selection-bar/selection-bar';
+import type { SelectionAction } from '@components/selection-bar/selection-bar';
 import { ViewLifecycleMixin } from '@utils/view-lifecycle';
 import { PHONE_QUERY } from '@utils/breakpoints';
 import {
@@ -77,7 +80,9 @@ import '@components/playlist-picker/playlist-picker.js';
 import type { TrackDetails } from '@components/track-details/track-details.js';
 import type { CoverArtUrls } from '@components/track-details/track-details.js';
 import {
+    ICON_PLAY,
     ICON_PLAYLIST,
+    ICON_PLAY_NEXT,
     ICON_QUEUE,
 } from '@utils/icon-language';
 
@@ -1304,6 +1309,8 @@ export class TrackList
             virt.removeEventListener('click', this.onDelegatedClick);
             virt.removeEventListener('dblclick', this.onDelegatedDblClick);
             virt.removeEventListener('contextmenu', this.onDelegatedContextMenu);
+            virt.removeEventListener('yj-tap', this.onRowTap);
+            virt.removeEventListener('yj-long-press', this.onRowLongPress);
             virt.removeEventListener('dragstart', this.onDelegatedDragStart);
             virt.removeEventListener('dragend', this.onTrackDragEnd);
         }
@@ -1445,6 +1452,11 @@ export class TrackList
         virt.addEventListener('contextmenu', this.onDelegatedContextMenu);
         virt.addEventListener('dragstart', this.onDelegatedDragStart);
         virt.addEventListener('dragend', this.onTrackDragEnd);
+        // Delegated like the rest: the gesture layer dispatches on the
+        // element the finger landed on, composed, so it arrives here
+        // through the same path a real click takes (plan 019).
+        virt.addEventListener('yj-tap', this.onRowTap);
+        virt.addEventListener('yj-long-press', this.onRowLongPress);
         this.delegationAttached = true;
     }
 
@@ -1708,6 +1720,62 @@ export class TrackList
         if (hit) this.onTrackContextMenu(e, hit.track);
     };
 
+    /**
+     * A finger tapped a row (plan 019, #63).
+     *
+     * On a desktop a click selects and a double-click plays; a finger
+     * inverts that, because there is no second button and no modifier
+     * key, so the primary action has to be the primary gesture.
+     *
+     * Claiming the gesture (`preventDefault`) is what tells the layer
+     * to swallow the click behind it -- otherwise playing a track
+     * would also select it, and the row would end up in both states.
+     * A tap this does *not* claim falls through as an ordinary click,
+     * which is what keeps the favourite icon working.
+     */
+    private onRowTap = (e: GestureEvent) => {
+        const hit = this.resolveTrackFromEvent(e);
+
+        if (!hit) return;
+
+        // A control inside the row owns its own tap. The same rule the
+        // shortcut service has for a focused control that owns a key,
+        // and without it the 44px favourite target (#56) becomes a
+        // 44px play target.
+        if ((e.target as HTMLElement).closest('.fav-icon')) return;
+
+        e.preventDefault();
+        this.focusedIndex = hit.index;
+
+        if (this.selection.selectionMode) {
+            this.selection.toggleInMode(hit.track.FilePath, hit.index);
+            this.virtualizer?.requestUpdate();
+
+            return;
+        }
+
+        this.playFromRow(hit.index);
+    };
+
+    /**
+     * A finger held a row still for half a second.
+     *
+     * Claiming this is what makes it *selection mode* rather than the
+     * context menu it has been since plan 016 -- an unclaimed
+     * `yj-long-press` still becomes a `contextmenu`, which is how the
+     * card grids and Explore keep the behaviour they have.
+     */
+    private onRowLongPress = (e: GestureEvent) => {
+        const hit = this.resolveTrackFromEvent(e);
+
+        if (!hit) return;
+
+        e.preventDefault();
+        this.focusedIndex = hit.index;
+        this.selection.enterSelectionMode(hit.track.FilePath, hit.index);
+        this.virtualizer?.requestUpdate();
+    };
+
     private onDelegatedDragStart = (e: DragEvent) => {
         const hit = this.resolveTrackFromEvent(e);
 
@@ -2238,6 +2306,45 @@ export class TrackList
         this.saveSortPreferences();
     };
 
+    /**
+     * The three worth a thumb. Everything else is behind "More",
+     * which opens the context menu this list already renders.
+     *
+     * A bar is one row on a 424px screen and the menu is nine items,
+     * so this is a subset by necessity rather than a second opinion
+     * about what matters -- and plan 018's promise that no action is
+     * unreachable is kept by the overflow, not by this list.
+     */
+    private static readonly SELECTION_ACTIONS: SelectionAction[] = [
+        { id: 'play', label: 'Play', icon: ICON_PLAY },
+        { id: 'add-to-queue', label: 'Add to queue', icon: ICON_QUEUE },
+        { id: 'play-next', label: 'Play next', icon: ICON_PLAY_NEXT },
+    ];
+
+    private renderSelectionBar() {
+        // Only in selection mode: a mouse selection is modeless and
+        // shows its actions on right-click, which is where a desktop
+        // user looks for them.
+        if (!this.selection.selectionMode) return nothing;
+
+        return html`
+            <selection-bar
+                .count=${this.selection.selectionCount}
+                .actions=${TrackList.SELECTION_ACTIONS}
+                @selection-exit=${this.onSelectionExit}
+                @selection-action=${(e: CustomEvent<{ id: string }>) =>
+                    this.onContextMenuAction(e.detail.id)}
+                @selection-more=${(e: CustomEvent<{ x: number; y: number }>) =>
+                    this.ctxMenu.openAt(e.detail.x, e.detail.y)}
+            ></selection-bar>
+        `;
+    }
+
+    private onSelectionExit = () => {
+        this.selection.exitSelectionMode();
+        this.virtualizer?.requestUpdate();
+    };
+
     override render() {
         const visibleTracks = this.cachedSortedTracks;
         const cols = this.activeColumns;
@@ -2322,6 +2429,7 @@ export class TrackList
                     )}
       </div>
             </div>
+            ${this.renderSelectionBar()}
           `}
 
       <menu-surface

frontend/src/styles/back-button.css.ts

@@ -0,0 +1,54 @@
+import { css } from 'lit';
+
+/**
+ * The way out of a detail view, at the app's 44px touch floor.
+ *
+ * #186's second table names `artist-details`' back button at
+ * **32x32**. It is the same declaration in **six** components —
+ * `artist-details`, `genre-details`, `playlist-details`,
+ * `smart-playlist-details`, `explore-artist-details` and
+ * `explore-album-details` — byte-identical, 32px in all six, and the
+ * sweep that filed the issue visited one of them.
+ *
+ * That is the argument for this file rather than six edits. A device
+ * sweep walks the views somebody thought to open, so six copies of a
+ * control is six chances for the next pass to miss five; the arrows
+ * and the toggles were each one declaration covering 36 and 29
+ * controls, and this is the same shape stated the other way round.
+ *
+ * **It is a real 44px box, not padding with the width handed back.**
+ * The header pass had to grow a hit area past its own layout box
+ * because `page-header` measures itself for #69's overflow fit; a
+ * detail view's header does not, so the control can simply be the
+ * target. It also *should* be — this button has a visible background,
+ * so a hit area larger than the circle would be a control that is
+ * bigger than it looks, which is the thing #187 accepts only where a
+ * thin painted track is the point.
+ *
+ * The size is #55's, arrived at for the same reason one component
+ * over: "the way out is 44px on a phone", when the queue panel's close
+ * button was 25x21 and, at phone width, the only pointer route off a
+ * full-screen surface. A detail view has the platform's back gesture
+ * as well, so this is less severe than the queue was — it is the same
+ * control wearing the same mistake.
+ */
+export const backButton = css`
+    .back-button {
+        display: flex;
+        align-items: center;
+        justify-content: center;
+        width: 44px;
+        height: 44px;
+        border: none;
+        border-radius: 50%;
+        background: var(--yj-bg-overlay, rgba(255, 255, 255, 0.06));
+        color: var(--yj-text-primary, #fff);
+        cursor: pointer;
+        flex-shrink: 0;
+        transition: background-color 0.15s ease;
+    }
+
+    .back-button:hover {
+        background: var(--yj-bg-hover, rgba(255, 255, 255, 0.12));
+    }
+`;

frontend/src/styles/wa-touch-floor.css.ts

@@ -0,0 +1,48 @@
+import { css } from 'lit';
+
+/**
+ * A Web Awesome form control is at least the app's 44px touch floor.
+ *
+ * #56 named 44px and #186 found nothing but the transport had reached
+ * it. Web Awesome's form controls are the part of Settings this app
+ * does not draw: measured on the reference device (TLP301, 424x439),
+ * `wa-input`'s control is **204x20** and `wa-button` **185x21** — the
+ * shortest controls on the page, and the only ones whose height is
+ * decided inside somebody else's shadow root.
+ *
+ * `--wa-form-control-height` is that decision, and it is the library's
+ * own theming variable rather than a part or an internal — the default
+ * theme sets it at `:root` and every control that has a height reads
+ * it (button, input, select, radio). So this is `wa-slider-label`'s
+ * better half: the API first, and no reach into a shadow root at all.
+ *
+ * Three things about it are load-bearing.
+ *
+ * **A custom property inherits through a shadow boundary**, which is
+ * what lets a `:host` declaration reach a `wa-input` the host renders.
+ * That is also why it is a stylesheet a component adopts rather than a
+ * `:root` rule in `index.css`: a `:root` rule would cover every wa
+ * control in the app in one line and be invisible to the component
+ * tier, which renders a component and no page stylesheet. Here the
+ * floor is measurable where it is applied.
+ *
+ * **It is a flat 44px rather than a floor over the library's own
+ * expression.** The default is `round(calc(2 * padding-block + 1em *
+ * line-height), 1px)` — em-based, so `size="small"` is what produced
+ * the 20px above — and a `max(44px, …)` would have to restate that
+ * formula here, which is a copy of somebody else's arithmetic that
+ * goes stale silently. A flat value is safe because this app uses
+ * exactly two sizes, `small` and the default, and both are under the
+ * floor; a `size="large"` added later would be pinned down to 44 and
+ * should take that as the prompt to revisit this.
+ *
+ * **Only the height is pinned.** The font size still comes from
+ * `size="small"`, so a control grows its hit area without growing its
+ * visual weight — which is what #186's Direction asks for and what the
+ * page header's second pass had to be corrected to do.
+ */
+export const waTouchFloor = css`
+    :host {
+        --wa-form-control-height: 44px;
+    }
+`;

frontend/src/utils/long-press.ts

@@ -1,201 +0,0 @@
-/**
- * Long-press as the touch equivalent of a right-click (plan 016 B2,
- * phase 3).
- *
- * Every context menu in the app opens from a `contextmenu` event —
- * `track-list` and `queue-panel` delegate one on their virtualizer,
- * the card grids and both playlist detail views bind one per row, and
- * `explore-artist-details` binds three. A phone has no right-click, so
- * a phone reached none of them.
- *
- * **This is one document listener, not six components' worth of touch
- * handling.** A press that stays still for `LONG_PRESS_MS` dispatches a
- * synthetic `contextmenu` at the touch point on the element the touch
- * actually landed on, and every existing handler — delegated or
- * per-row, in any shadow root — runs unchanged. Six implementations of
- * a gesture is exactly the fault `ContextMenuController` exists to
- * prevent, and a seam that needs no component to opt in cannot be
- * forgotten by the next component.
- *
- * Three things about it are load-bearing.
- *
- * **The target comes from `composedPath()[0]`, not from
- * `elementFromPoint`**, which stops at the outermost shadow host: every
- * menu in this app is bound inside one, so a synthetic event dispatched
- * on the host reaches a delegated listener and no per-row one.
- *
- * **A browser that already does this must win.** Chromium fires a
- * `contextmenu` on long-press itself; WebKitGTK and the Android WebView
- * vary. So one arriving during the press cancels ours, and one arriving
- * just after ours is swallowed at document capture — where nothing else
- * has seen it yet. The two are told apart by **identity** (a `WeakSet`
- * of the events this module made) rather than by `isTrusted`, so the
- * suppressor cannot eat the event it exists to deliver, the rule holds
- * for anything else in the app that synthesises one, and a test can
- * stand in for a browser that fires its own.
- *
- * **The click that ends the gesture is swallowed.** A row's click
- * selects, and a card's plays; without this, opening a menu also
- * activates the thing under it. It is keyed on the gesture (cleared by
- * the next `pointerdown`) rather than on a time window, so a quick tap
- * on the menu that just opened is not eaten too.
- */
-
-/** How long a press must hold still to mean "menu". */
-export const LONG_PRESS_MS = 500;
-
-/**
- * How far a press may drift and still count. Below a finger's own
- * jitter is a gesture nobody can perform; above ~12px it starts
- * stealing the first frames of a scroll.
- */
-export const MOVE_TOLERANCE_PX = 10;
-
-/** The active installation, so a second call is a no-op rather than a
- *  second listener set. */
-let uninstall: (() => void) | null = null;
-
-/** The events this module dispatched. Identity, not `isTrusted`: see
- *  the note above. */
-const ours = new WeakSet<Event>();
-
-/**
- * Install the gesture. Idempotent; returns the uninstaller (which the
- * tests use — the app installs once and never removes it).
- */
-export function installLongPressContextMenu(): () => void {
-    if (uninstall) return uninstall;
-
-    let timer: ReturnType<typeof setTimeout> | null = null;
-    let originX = 0;
-    let originY = 0;
-    let target: EventTarget | null = null;
-
-    /** A trusted `contextmenu` arrived for this press: the browser has
-     *  it covered. */
-    let nativeSeen = false;
-
-    /** We opened a menu, and the click ending that gesture is not a
-     *  click on anything. */
-    let swallowClick = false;
-
-    /** We dispatched one, so a trusted one arriving now is a duplicate. */
-    let justFired = false;
-
-    const cancel = (): void => {
-        if (timer !== null) clearTimeout(timer);
-
-        timer = null;
-        target = null;
-    };
-
-    const fire = (): void => {
-        timer = null;
-
-        const el = target;
-
-        target = null;
-
-        if (nativeSeen || !el) return;
-
-        justFired = true;
-        swallowClick = true;
-
-        const menu = new MouseEvent('contextmenu', {
-            bubbles: true,
-            cancelable: true,
-            // Or it stops at the shadow root the row lives in, and the
-            // delegated listeners never see it.
-            composed: true,
-            clientX: originX,
-            clientY: originY,
-            button: 2,
-        });
-
-        ours.add(menu);
-        el.dispatchEvent(menu);
-    };
-
-    const onPointerDown = (e: PointerEvent): void => {
-        // A new gesture: whatever the last one left behind is stale.
-        swallowClick = false;
-        justFired = false;
-        nativeSeen = false;
-        cancel();
-
-        if (e.pointerType !== 'touch' || !e.isPrimary) return;
-
-        originX = e.clientX;
-        originY = e.clientY;
-        target = e.composedPath()[0] ?? e.target;
-        timer = setTimeout(fire, LONG_PRESS_MS);
-    };
-
-    const onPointerMove = (e: PointerEvent): void => {
-        if (timer === null) return;
-
-        const drifted =
-            Math.abs(e.clientX - originX) > MOVE_TOLERANCE_PX ||
-            Math.abs(e.clientY - originY) > MOVE_TOLERANCE_PX;
-
-        if (drifted) cancel();
-    };
-
-    const onContextMenu = (e: Event): void => {
-        // Ours. Everything below is about somebody else's.
-        if (ours.has(e)) return;
-
-        if (timer !== null) {
-            // The browser got there first, so stand down rather than
-            // opening the same menu twice.
-            nativeSeen = true;
-            cancel();
-
-            return;
-        }
-
-        if (justFired) {
-            justFired = false;
-            e.preventDefault();
-            e.stopImmediatePropagation();
-        }
-    };
-
-    const onClick = (e: Event): void => {
-        if (!swallowClick) return;
-
-        swallowClick = false;
-        e.preventDefault();
-        e.stopImmediatePropagation();
-    };
-
-    // Capture throughout: a component handler that stops propagation
-    // (every context-menu handler in the app does) must not be able to
-    // hide the gesture from this, and the suppressors have to run
-    // before anything that would act on the event.
-    const opts = { capture: true } as const;
-
-    document.addEventListener('pointerdown', onPointerDown, opts);
-    document.addEventListener('pointermove', onPointerMove, opts);
-    document.addEventListener('pointerup', cancel, opts);
-    document.addEventListener('pointercancel', cancel, opts);
-    document.addEventListener('contextmenu', onContextMenu, opts);
-    document.addEventListener('click', onClick, opts);
-    // A scroll started by something other than the finger (momentum, a
-    // programmatic reveal) still means the press was not a press.
-    document.addEventListener('scroll', cancel, { capture: true, passive: true });
-
-    uninstall = () => {
-        cancel();
-        document.removeEventListener('pointerdown', onPointerDown, opts);
-        document.removeEventListener('pointermove', onPointerMove, opts);
-        document.removeEventListener('pointerup', cancel, opts);
-        document.removeEventListener('pointercancel', cancel, opts);
-        document.removeEventListener('contextmenu', onContextMenu, opts);
-        document.removeEventListener('click', onClick, opts);
-        document.removeEventListener('scroll', cancel, opts);
-        uninstall = null;
-    };
-
-    return uninstall;
-}

frontend/src/utils/selection-controller.ts

@@ -19,6 +19,29 @@ export class SelectionController implements ReactiveController {
     private host: SelectionHost;
     private _selectedItems: Set<string> = new Set();
     private lastSelectedIndex: number | null = null;
+    private _mode = false;
+
+    /**
+     * Whether the list is in *selection mode* (plan 019, #63).
+     *
+     * A finger has no modifier keys, so the ctrl/shift semantics this
+     * controller was written for cannot be expressed by touch at all.
+     * Selection mode is the platform's answer: a long press enters it,
+     * and while it is on, a tap toggles a row instead of playing it.
+     *
+     * It is a flag *here* rather than a fifth concept beside the
+     * controller because all four surfaces that select
+     * (`track-list`, `queue-panel` and both playlist detail views)
+     * already share this class -- so "is this list selecting" has one
+     * answer per list, in the object that already owns the selection
+     * it would otherwise contradict.
+     *
+     * A mouse never sets it. Desktop selection is unchanged and stays
+     * modeless, which is what `handleItemClick` still implements.
+     */
+    get selectionMode(): boolean {
+        return this._mode;
+    }
 
     constructor(host: SelectionHost) {
         this.host = host;
@@ -136,8 +159,64 @@ export class SelectionController implements ReactiveController {
         return true;
     }
 
+    /**
+     * Enter selection mode with `key` selected.
+     *
+     * The row the gesture was made on is selected, rather than the
+     * mode opening empty: a long press is a statement about *that*
+     * row, and an action bar with nothing in it is a mode the user has
+     * to make a second gesture to escape.
+     */
+    enterSelectionMode(key: string, index: number): void {
+        this._mode = true;
+        this._selectedItems = new Set([key]);
+        this.lastSelectedIndex = index;
+        this.host.requestUpdate();
+        this.host.onSelectionChanged?.();
+    }
+
+    /**
+     * Toggle one row, and leave the mode when the last one goes.
+     *
+     * Deselecting everything is how Android's own list surfaces exit
+     * selection mode, and it matters more here than convention: the
+     * mode changes what a tap *means*, so a mode with an empty
+     * selection is a list where tapping does nothing and nothing on
+     * screen says why.
+     */
+    toggleInMode(key: string, index: number): void {
+        const next = new Set(this._selectedItems);
+
+        if (next.has(key)) next.delete(key);
+        else next.add(key);
+
+        this._selectedItems = next;
+        this.lastSelectedIndex = index;
+
+        if (next.size === 0) this._mode = false;
+
+        this.host.requestUpdate();
+        this.host.onSelectionChanged?.();
+    }
+
+    /** Leave selection mode, dropping the selection with it. */
+    exitSelectionMode(): void {
+        if (!this._mode && this._selectedItems.size === 0) return;
+
+        this._mode = false;
+        this._selectedItems = new Set();
+        this.lastSelectedIndex = null;
+        this.host.requestUpdate();
+        this.host.onSelectionChanged?.();
+    }
+
     /** Clear the entire selection. */
     clear(): void {
+        // The mode goes with it: every caller of this means "the
+        // selection is no longer meaningful", and a mode outliving the
+        // selection it was showing is the empty-mode trap above.
+        this._mode = false;
+
         if (this._selectedItems.size === 0) return;
 
         this._selectedItems = new Set();
@@ -176,6 +255,13 @@ export class SelectionController implements ReactiveController {
         if (next.size === this._selectedItems.size) return;
 
         this._selectedItems = next;
+
+        // A refetch that emptied the selection also ends the mode --
+        // otherwise removing the last selected track from the library
+        // leaves the list in a state where a tap selects and the bar
+        // is gone.
+        if (next.size === 0) this._mode = false;
+
         this.host.requestUpdate();
         this.host.onSelectionChanged?.();
     }

frontend/src/utils/touch-gestures.ts

@@ -0,0 +1,327 @@
+/**
+ * The touch gestures, as one document listener (plan 019, #63).
+ *
+ * This replaces `utils/long-press.ts` rather than sitting beside it,
+ * and that is the point: two document listeners both claiming the
+ * 500ms hold is exactly the fault that file's own header warns about.
+ * What it did — one capture listener, the target from
+ * `composedPath()[0]`, the browser's own gesture winning, the trailing
+ * click swallowed — is kept whole. What changes is what the gesture
+ * *means*.
+ *
+ * **A gesture is announced, not acted on.** Two composed, cancelable
+ * events are dispatched on the element the finger actually landed on:
+ *
+ *   `yj-tap`         a short press that did not drift
+ *   `yj-long-press`  a press that held still for LONG_PRESS_MS
+ *
+ * A component that wants the gesture handles it and calls
+ * `preventDefault()`. Nothing else changes. That shape is what lets
+ * this reassign the hold without touching a single one of the fourteen
+ * context menus downstream of it: **an unclaimed `yj-long-press` still
+ * becomes a synthetic `contextmenu`**, so a card grid, an Explore
+ * result or a playlist row behaves exactly as it did, and only the
+ * lists that opt in get selection mode.
+ *
+ * The same rule keeps taps honest. An unclaimed `yj-tap` does nothing
+ * at all and the browser's click follows normally, so every button,
+ * link and checkbox in the app is untouched by this file. Only a
+ * claimed tap has its click swallowed — otherwise playing a track
+ * would also select it.
+ *
+ * Five things are load-bearing.
+ *
+ * **The predicate is the pointer, not the platform** (plan 019,
+ * decision 1). `pointerType === 'touch'`, per event — so an Android
+ * tablet over 600px, a touchscreen laptop with a mouse also plugged
+ * in, and a narrow desktop window are all right for free, and there is
+ * no second declaration of what a phone does. Keyed on a viewport
+ * width, the first of those three gets desktop semantics on a
+ * touchscreen, which is the inversion #63 exists to fix, on the
+ * platform it exists for.
+ *
+ * **There is no double-tap**, and it is not an omission — see plan
+ * 019, decision 2. Measured on the reference device, the play command
+ * to `TrackChanged` is ~100ms; a double-tap discriminator has to hold
+ * every tap for the app's own `DOUBLE_CLICK_GRACE_MS` of 250 before it
+ * can act, which is 3.5x the primary interaction in the app to reach a
+ * menu that long-press already reaches.
+ *
+ * **The target comes from `composedPath()[0]`**, not
+ * `elementFromPoint`, which stops at the outermost shadow host: every
+ * list in this app delegates inside one, so an event dispatched on the
+ * host reaches a delegated listener and no per-row one.
+ *
+ * **A browser that fires its own `contextmenu` is a trigger, not a
+ * competitor**, and that is a change from `long-press.ts` rather than
+ * an inherited rule. It used to stand down when a trusted
+ * `contextmenu` arrived, because both paths ended in the same place: a
+ * context menu. They no longer do — ours may end in selection mode —
+ * so standing down means the gesture silently does the *old* thing.
+ *
+ * Measured on the reference device, which is the only tier that can
+ * see this: Chrome 113's WebView fires its own `contextmenu` on a long
+ * press, so a hold on a track row opened the context menu and
+ * `yj-long-press` was never announced at all. Every test in the
+ * component tier passed, because dispatched pointer events do not make
+ * a browser synthesise one.
+ *
+ * So a trusted `contextmenu` arriving mid-press *becomes* the long
+ * press: `yj-long-press` is announced from it, and only if a component
+ * claims it is the native event suppressed. Unclaimed, it propagates
+ * untouched and opens the menu it always did — which is the same
+ * "browser wins" outcome, now reached by asking rather than assuming.
+ *
+ * Ours and the browser's are still told apart by identity rather than
+ * `isTrusted` — a `WeakSet` of the events this module made — so the
+ * suppressor cannot eat the event it exists to deliver, and a test can
+ * stand in for a browser that fires one.
+ *
+ * **The click swallow is keyed on the gesture**, cleared by the next
+ * `pointerdown` rather than by a time window, so the first tap on a
+ * sheet that just opened is not eaten too.
+ */
+
+/** How long a press must hold still to mean "long press". */
+export const LONG_PRESS_MS = 500;
+
+/**
+ * How far a press may drift and still count. Below a finger's own
+ * jitter is a gesture nobody can perform; above ~12px it starts
+ * stealing the first frames of a scroll.
+ */
+export const MOVE_TOLERANCE_PX = 10;
+
+/** Detail carried by both gesture events. */
+export interface GestureDetail {
+    /** Where the finger was, in client coordinates — a menu opens here. */
+    x: number;
+    y: number;
+}
+
+export type GestureEvent = CustomEvent<GestureDetail>;
+
+declare global {
+    interface HTMLElementEventMap {
+        'yj-tap': GestureEvent;
+        'yj-long-press': GestureEvent;
+    }
+}
+
+/** The active installation, so a second call is a no-op rather than a
+ *  second listener set. */
+let uninstall: (() => void) | null = null;
+
+/** The events this module dispatched. Identity, not `isTrusted`. */
+const ours = new WeakSet<Event>();
+
+/**
+ * Install the gestures. Idempotent; returns the uninstaller (which the
+ * tests use — the app installs once and never removes it).
+ */
+export function installTouchGestures(): () => void {
+    if (uninstall) return uninstall;
+
+    let timer: ReturnType<typeof setTimeout> | null = null;
+    let originX = 0;
+    let originY = 0;
+    let target: EventTarget | null = null;
+
+    /** The press is still a candidate for a tap: it has neither
+     *  drifted nor become a long press. */
+    let tapCandidate = false;
+
+    /** A trusted `contextmenu` arrived for this press: the browser has
+     *  it covered. */
+    let nativeSeen = false;
+
+    /** A gesture was claimed, and the click ending it is not a click on
+     *  anything. */
+    let swallowClick = false;
+
+    /** We dispatched a `contextmenu`, so a trusted one arriving now is
+     *  a duplicate. */
+    let justFired = false;
+
+    const cancel = (): void => {
+        if (timer !== null) clearTimeout(timer);
+
+        timer = null;
+        target = null;
+        tapCandidate = false;
+    };
+
+    /**
+     * Announce a gesture on the element the finger landed on.
+     * Returns whether a component claimed it.
+     */
+    const announce = (name: 'yj-tap' | 'yj-long-press', el: EventTarget): boolean => {
+        const event: GestureEvent = new CustomEvent<GestureDetail>(name, {
+            bubbles: true,
+            cancelable: true,
+            // Or it stops at the shadow root the row lives in, and the
+            // delegated listeners never see it.
+            composed: true,
+            detail: { x: originX, y: originY },
+        });
+
+        ours.add(event);
+        el.dispatchEvent(event);
+
+        return event.defaultPrevented;
+    };
+
+    const fireContextMenu = (el: EventTarget): void => {
+        justFired = true;
+
+        const menu = new MouseEvent('contextmenu', {
+            bubbles: true,
+            cancelable: true,
+            composed: true,
+            clientX: originX,
+            clientY: originY,
+            button: 2,
+        });
+
+        ours.add(menu);
+        el.dispatchEvent(menu);
+    };
+
+    const onLongPress = (): void => {
+        timer = null;
+        tapCandidate = false;
+
+        const el = target;
+
+        target = null;
+
+        if (nativeSeen || !el) return;
+
+        // The gesture happened either way, so the click that ends it is
+        // never a click on anything -- whether a list claimed it for
+        // selection mode or a card grid let it fall through to a menu.
+        swallowClick = true;
+
+        // An unclaimed long press is what it has always been. This is
+        // the whole reason the fourteen context menus need no change.
+        if (!announce('yj-long-press', el)) fireContextMenu(el);
+    };
+
+    const onPointerDown = (e: PointerEvent): void => {
+        // A new gesture: whatever the last one left behind is stale.
+        swallowClick = false;
+        justFired = false;
+        nativeSeen = false;
+        cancel();
+
+        if (e.pointerType !== 'touch' || !e.isPrimary) return;
+
+        originX = e.clientX;
+        originY = e.clientY;
+        target = e.composedPath()[0] ?? e.target;
+        tapCandidate = true;
+        timer = setTimeout(onLongPress, LONG_PRESS_MS);
+    };
+
+    const onPointerMove = (e: PointerEvent): void => {
+        if (timer === null) return;
+
+        const drifted =
+            Math.abs(e.clientX - originX) > MOVE_TOLERANCE_PX ||
+            Math.abs(e.clientY - originY) > MOVE_TOLERANCE_PX;
+
+        // A drifted press is neither gesture -- it is a scroll, and the
+        // virtualizer's, not ours.
+        if (drifted) cancel();
+    };
+
+    const onPointerUp = (): void => {
+        const el = target;
+        const wasTap = tapCandidate && timer !== null;
+
+        // Clears the long-press timer, so a tap cannot also become one.
+        cancel();
+
+        if (!wasTap || !el) return;
+
+        // Only a *claimed* tap swallows its click. An unclaimed one has
+        // to fall through untouched, or every button in the app stops
+        // working.
+        if (announce('yj-tap', el)) swallowClick = true;
+    };
+
+    const onContextMenu = (e: Event): void => {
+        // Ours. Everything below is about somebody else's.
+        if (ours.has(e)) return;
+
+        if (timer !== null) {
+            // The browser recognised the same hold this was timing.
+            // Use its event as the trigger rather than racing it --
+            // and rather than standing down, which is what the old
+            // rule did and which now silently means "do the thing this
+            // gesture used to do".
+            const el = e.composedPath()[0] ?? e.target;
+
+            nativeSeen = true;
+            cancel();
+
+            if (!el) return;
+
+            swallowClick = true;
+
+            // Claimed: the component wants selection mode, so the
+            // browser's menu must not also open. Unclaimed: let it
+            // through exactly as before.
+            if (announce('yj-long-press', el)) {
+                e.preventDefault();
+                e.stopImmediatePropagation();
+            }
+
+            return;
+        }
+
+        if (justFired) {
+            justFired = false;
+            e.preventDefault();
+            e.stopImmediatePropagation();
+        }
+    };
+
+    const onClick = (e: Event): void => {
+        if (!swallowClick) return;
+
+        swallowClick = false;
+        e.preventDefault();
+        e.stopImmediatePropagation();
+    };
+
+    // Capture throughout: a component handler that stops propagation
+    // (every context-menu handler in the app does) must not be able to
+    // hide the gesture from this, and the suppressors have to run
+    // before anything that would act on the event.
+    const opts = { capture: true } as const;
+
+    document.addEventListener('pointerdown', onPointerDown, opts);
+    document.addEventListener('pointermove', onPointerMove, opts);
+    document.addEventListener('pointerup', onPointerUp, opts);
+    document.addEventListener('pointercancel', cancel, opts);
+    document.addEventListener('contextmenu', onContextMenu, opts);
+    document.addEventListener('click', onClick, opts);
+    // A scroll started by something other than the finger (momentum, a
+    // programmatic reveal) still means the press was not a press.
+    document.addEventListener('scroll', cancel, { capture: true, passive: true });
+
+    uninstall = () => {
+        cancel();
+        document.removeEventListener('pointerdown', onPointerDown, opts);
+        document.removeEventListener('pointermove', onPointerMove, opts);
+        document.removeEventListener('pointerup', onPointerUp, opts);
+        document.removeEventListener('pointercancel', cancel, opts);
+        document.removeEventListener('contextmenu', onContextMenu, opts);
+        document.removeEventListener('click', onClick, opts);
+        document.removeEventListener('scroll', cancel, opts);
+        uninstall = null;
+    };
+
+    return uninstall;
+}

frontend/test/components/detail-touch-targets.test.ts

@@ -0,0 +1,230 @@
+/**
+ * The controls #186's second table found, outside Settings.
+ *
+ * Six one-off controls across four surfaces, and the reason they are a
+ * test rather than six stylesheet edits is `back-button`. The issue
+ * names it in `artist-details` at **32x32**; it is the same
+ * declaration, byte-identical, in *six* components — because a device
+ * sweep walks the views somebody thought to open, and five of them
+ * were not opened.
+ *
+ * So the assertion is over the whole set rather than over the one that
+ * was measured. That is `icon-language.test.ts`'s shape and it is here
+ * for the same reason: checking one call site checks one call site.
+ *
+ * | control | before | where |
+ * |---|---|---|
+ * | `.folders-menu-trigger` | **32x18** | autotag |
+ * | `.section-toggle` | 187x**15** | autotag |
+ * | `.back-button` | 32x32 | six detail views |
+ * | Requests / Downloads tabs | 85x**34**, 96x**34** | downloads |
+ * | `.search-mode-tab` | 89x**26**, 79x**26** | explore |
+ * | explore search input | 325x**18** in a 36px box | explore |
+ *
+ * `page-action-check-now` (113x29) is in that table and is **not**
+ * here: it is a `PageAction`, so #195 raised it with the rest of the
+ * page header's actions, and `touch-targets.test.ts` already covers
+ * it. Re-asserting it here would be a second statement of one rule.
+ */
+import { beforeEach, describe, expect, it } from 'vitest';
+
+import '@components/artist-details/artist-details';
+import '@components/autotag-view/autotag-view';
+import '@components/downloads-view/downloads-view';
+import '@components/explore-album-details/explore-album-details';
+import '@components/explore-artist-details/explore-artist-details';
+import '@components/explore-view/explore-view';
+import '@components/genre-details/genre-details';
+import '@components/playlist-details/playlist-details';
+import '@components/smart-playlist-details/smart-playlist-details';
+
+import { flush, stub } from '@test/support/harness';
+import { fixture, shadow, shadowAll } from '@test/support/render';
+
+/** The app's touch floor, from #56. */
+const FLOOR = 44;
+
+/**
+ * Every component that draws a back button.
+ *
+ * The list is here rather than derived because deriving it means
+ * reading the source, and this tier renders instead — but it is
+ * checked against the source by `the back button is one declaration`
+ * below, so a seventh view cannot join quietly.
+ */
+const BACK_BUTTON_VIEWS = [
+  'artist-details',
+  'genre-details',
+  'playlist-details',
+  'smart-playlist-details',
+  'explore-artist-details',
+  'explore-album-details',
+] as const;
+
+function boxOf(el: Element | null | undefined): { w: number; h: number } {
+  if (!el) return { w: 0, h: 0 };
+
+  const box = el.getBoundingClientRect();
+
+  return { w: Math.round(box.width), h: Math.round(box.height) };
+}
+
+describe('the way out of a detail view', () => {
+  beforeEach(() => {
+    for (const path of [
+      'library.Library.GetTracks',
+      'library.Library.GetAlbums',
+      'library.Library.GetArtists',
+      'library.Library.GetGenres',
+      'playlist.Service.GetAllPlaylists',
+      'playlist.Service.GetAllPlaylistsWithTracks',
+    ]) {
+      stub(path, []);
+    }
+  });
+
+  it.each(BACK_BUTTON_VIEWS)('is 44px in <%s>', async (tag) => {
+    // #55 settled this one component over, when the queue panel's
+    // close button was 25x21 and, at phone width, the only pointer
+    // route off a full-screen surface: "the way out is 44px". A detail
+    // view has the platform's back gesture as well, so it is less
+    // severe -- and it is the same control wearing the same mistake.
+    const el = await fixture(tag);
+
+    await flush();
+
+    const back = shadow(el, '.back-button');
+
+    expect(back, `${tag} draws a back button`).toBeTruthy();
+    expect(boxOf(back)).toEqual({ w: FLOOR, h: FLOOR });
+  });
+
+  it('is one declaration, so a seventh view cannot miss it', async () => {
+    // The regression this exists for is not a size changing -- it is
+    // somebody adding a detail view and writing `.back-button` out
+    // again at 32px, which is exactly how there came to be six copies.
+    // A sweep of the running app would not catch it either, because a
+    // sweep visits the views you think to open.
+    const sources = import.meta.glob('../../src/components/**/*.ts', {
+      query: '?raw',
+      import: 'default',
+      eager: true,
+    }) as Record<string, string>;
+
+    expect(Object.keys(sources).length, 'the glob read something').toBeGreaterThan(0);
+
+    const redeclared = Object.entries(sources)
+      .filter(([, src]) => /^\s*\.back-button\s*(?::[a-z-]+\s*)?\{/m.test(src))
+      .map(([path]) => path);
+
+    expect(redeclared).toEqual([]);
+  });
+});
+
+describe('autotag', () => {
+  it('raises the two smallest controls the sweep found', async () => {
+    // 187x15 and 32x18. The section toggle was the smallest control
+    // measured anywhere in the app until the column arrows were
+    // counted, and autotag is off by default (#25), which is
+    // presumably why nobody had met either.
+    const el = await fixture('autotag-view');
+
+    await flush();
+
+    for (const selector of ['.section-toggle', '.folders-menu-trigger']) {
+      const control = shadowAll<HTMLElement>(el, selector).find(
+        (c) => c.getBoundingClientRect().height > 0,
+      );
+
+      if (!control) continue;
+
+      expect(boxOf(control).h, `${selector} height`).toBeGreaterThanOrEqual(FLOOR);
+    }
+
+    // The stylesheet is the assertion for whichever of the two this
+    // fixture does not render -- both are behind state a bare mount
+    // does not reach, and a test that silently checked nothing is the
+    // trap icon-language.test.ts's first assertion exists for.
+    const sheet = (el.constructor as typeof HTMLElement & { styles?: unknown })
+      .styles;
+
+    expect(String(sheet)).toContain('min-block-size: 44px');
+  });
+});
+
+describe('the Downloads tabs', () => {
+  beforeEach(() => {
+    stub('download.Service.ListDownloads', []);
+    stub('download.Service.ListRequests', []);
+    stub('download.Service.ListProviders', []);
+  });
+
+  it('are the only route to their panels, and are 44px', async () => {
+    const el = await fixture('downloads-view');
+
+    await flush();
+
+    const tabs = shadowAll<HTMLElement>(el, '[role="tab"]');
+
+    expect(tabs).toHaveLength(2);
+
+    for (const tab of tabs) {
+      expect(boxOf(tab).h, tab.textContent?.trim()).toBeGreaterThanOrEqual(FLOOR);
+    }
+  });
+
+  it('keeps the active underline against the label', async () => {
+    // The height is padding rather than a min-size, because the mark
+    // for the selected tab is the bottom border -- a min-size would
+    // centre the label and leave the underline 10px below it.
+    const el = await fixture('downloads-view');
+
+    await flush();
+
+    const tab = shadowAll<HTMLElement>(el, '[role="tab"]')[0]!;
+    const style = getComputedStyle(tab);
+
+    expect(parseFloat(style.paddingBlockStart)).toBeGreaterThan(8);
+    expect(style.paddingBlockStart).toBe(style.paddingBlockEnd);
+  });
+});
+
+describe("Explore's own search row", () => {
+  beforeEach(() => {
+    stub('explore.Service.GetShelves', { State: 'ready', Shelves: [] });
+    stub('explore.Service.GetIndexStatus', {});
+  });
+
+  it('raises the mode tabs', async () => {
+    const el = await fixture('explore-view');
+
+    await flush();
+
+    const tabs = shadowAll<HTMLElement>(el, '.search-mode-tab');
+
+    expect(tabs.length).toBeGreaterThan(0);
+
+    for (const tab of tabs) {
+      expect(boxOf(tab).h, tab.textContent?.trim()).toBeGreaterThanOrEqual(FLOOR);
+    }
+  });
+
+  it('makes the whole search box the input, not the middle 18px of it', async () => {
+    // Two faults, not one: the row was 36px and the input inside it
+    // was **18**, so half the box was not a target at all -- a tap
+    // near the top or bottom edge landed on the container and did
+    // nothing. The container is 44 and the input stretches to fill it.
+    const el = await fixture('explore-view');
+
+    await flush();
+
+    const box = shadow(el, '.search-container');
+    const input = shadow(el, '.search-container input');
+
+    expect(box, 'the search row renders').toBeTruthy();
+    expect(input, 'it holds an input').toBeTruthy();
+
+    expect(boxOf(box).h).toBeGreaterThanOrEqual(FLOOR);
+    expect(boxOf(input).h).toBeGreaterThanOrEqual(FLOOR);
+  });
+});

frontend/test/components/hover-affordance.test.ts

@@ -1,5 +1,6 @@
 /**
- * A hover affordance is gated on the device having hover.
+ * A hover affordance is gated on the device having hover — in whichever
+ * direction keeps the action reachable.
  *
  * The home page's cover cards reveal a play button on :hover. A touch
  * long-press synthesises a hover state in the WebView, so on a phone
@@ -7,6 +8,15 @@
  * utils/long-press.ts is measuring for a context menu — a control
  * appearing because the user was reaching for a different one.
  *
+ * #137 is the same sweep with the opposite answer for two of its three
+ * cases. Where the revealed control is the *only* route to its action,
+ * hiding it removes the action, so it is always visible where there is
+ * no hover: `track-details`'s cover-art overlay and remove, and
+ * `shortcut-capture`'s reset. The queue's per-row remove is the third,
+ * and is the redundant kind — since #60 the row's context menu is a
+ * bottom sheet carrying "Remove from Queue" — so it takes #68's
+ * treatment here.
+ *
  * This is asserted against the *parsed stylesheet* rather than by
  * emulating a touch device, and that is a limitation worth stating
  * rather than hiding. CDP's Emulation.setEmulatedMedia does not reach
@@ -24,6 +34,9 @@
 import { describe, expect, it } from 'vitest';
 
 import '@components/home-view/home-view';
+import '@components/queue-panel/queue-panel';
+import '@components/track-details/track-details';
+import '@components/config-page/shortcut-capture';
 import { fixture } from '@test/support/render';
 
 /** Every rule in the element's own adopted stylesheets, flattened. */
@@ -83,3 +96,91 @@ describe('the home card play button', () => {
     expect(unconditional.some((r) => /display:\s*none/.test(r.text))).toBe(true);
   });
 });
+
+describe("the queue row's remove button", () => {
+  it('is absent where the device has no hover, the menu carrying the action', async () => {
+    const el = await fixture('queue-panel', {});
+    const rules = rulesOf(el);
+
+    expect(rules.length).toBeGreaterThan(0);
+
+    // visibility:hidden alone would leave an invisible button holding
+    // its hit area on a phone, which is the trap #68's commit names.
+    const unconditional = rules.filter(
+      (r) => r.condition === null && r.text.startsWith('.remove-button'),
+    );
+
+    expect(unconditional.length).toBeGreaterThan(0);
+    expect(unconditional.some((r) => /display:\s*none/.test(r.text))).toBe(true);
+
+    const reveals = rules.filter(
+      (r) =>
+        r.text.includes('.remove-button') && /visibility:\s*visible/.test(r.text),
+    );
+
+    expect(reveals.length).toBeGreaterThan(0);
+
+    for (const rule of reveals) {
+      expect(rule.condition).toMatch(/hover:\s*hover/);
+      expect(rule.condition).toMatch(/pointer:\s*fine/);
+    }
+  });
+});
+
+/**
+ * The two affordances that are the only route to their action.
+ *
+ * Asserted as "there is a rule showing it, and its condition is a
+ * *negated* hover query" — the same stylesheet reading as above, for
+ * the same reason: this tier's iframe cannot be emulated as a touch
+ * device, and the regression worth catching is someone folding the rule
+ * away as redundant on the desktop it does nothing on.
+ */
+describe('an affordance with no other route', () => {
+  const cases: Array<[string, string, string[]]> = [
+    ['track-details', 'track-details', ['.cover-art-overlay', '.cover-art-remove']],
+    ['shortcut-capture', 'shortcut-capture', ['.reset-btn']],
+  ];
+
+  for (const [name, tag, selectors] of cases) {
+    it(`${name} shows it where the device has no hover`, async () => {
+      const el = await fixture(tag, {});
+      const rules = rulesOf(el);
+
+      expect(rules.length).toBeGreaterThan(0);
+
+      for (const selector of selectors) {
+        const shown = rules.filter(
+          (r) =>
+            r.condition !== null &&
+            r.text.includes(selector) &&
+            /opacity:\s*1/.test(r.text),
+        );
+
+        const touch = shown.filter((r) => /not[\s\S]*hover:\s*hover/.test(r.condition!));
+
+        expect(touch.length).toBeGreaterThan(0);
+      }
+    });
+  }
+
+  // The one half this tier can measure rather than read: the query is
+  // negated, so on the hover-capable browser running these tests the
+  // control must still be revealed by hover and by nothing else. A rule
+  // written without the `not` would show it here, permanently, on every
+  // desktop.
+  it('leaves the desktop reveal alone, where the device does have hover', async () => {
+    expect(matchMedia('(hover: hover)').matches).toBe(true);
+
+    const el = await fixture('shortcut-capture', {
+      action: 'player.next',
+      label: 'Next Track',
+      currentKey: 'X',
+      defaultKey: 'N',
+    });
+    const btn = el.shadowRoot?.querySelector('.reset-btn');
+
+    expect(btn).not.toBeNull();
+    expect(getComputedStyle(btn!).opacity).toBe('0');
+  });
+});

frontend/test/components/long-press.test.ts

@@ -1,212 +0,0 @@
-/**
- * Long-press as the touch route to a context menu (plan 016 B2 phase 3).
- *
- * These run in a real browser with real event dispatch, which is the
- * only place the two things that make this hard are true: the synthetic
- * event has to cross a shadow boundary to reach the listener a
- * component actually bound, and the suppressors have to tell a trusted
- * event from ours at document capture without eating the one they exist
- * to deliver.
- *
- * The timings are real rather than faked, because the thing under test
- * *is* a timing, and 600 ms twice is cheaper than a fake-timer harness
- * that would also have to fake the pointer events.
- */
-import { describe, expect, it, afterEach, beforeEach } from 'vitest';
-
-import {
-  installLongPressContextMenu,
-  LONG_PRESS_MS,
-  MOVE_TOLERANCE_PX,
-} from '@utils/long-press';
-
-/** A press that has certainly resolved, either way. */
-const HELD = LONG_PRESS_MS + 120;
-
-/** A press that has certainly not. */
-const BRIEF = Math.round(LONG_PRESS_MS / 4);
-
-const wait = (ms: number) => new Promise((r) => setTimeout(r, ms));
-
-let uninstall: (() => void) | null = null;
-let host: HTMLElement;
-let inner: HTMLElement;
-
-/** A row inside a shadow root, which is where every menu in this app
- *  is bound — an element in the light DOM would pass a weaker test. */
-function mountRow(): { host: HTMLElement; inner: HTMLElement } {
-  const el = document.createElement('div');
-  const root = el.attachShadow({ mode: 'open' });
-  const row = document.createElement('div');
-
-  row.textContent = 'a track';
-  root.append(row);
-  document.body.append(el);
-
-  return { host: el, inner: row };
-}
-
-function press(
-  el: EventTarget,
-  type: string,
-  init: PointerEventInit = {},
-): void {
-  el.dispatchEvent(
-    new PointerEvent(type, {
-      bubbles: true,
-      composed: true,
-      cancelable: true,
-      pointerType: 'touch',
-      isPrimary: true,
-      clientX: 40,
-      clientY: 60,
-      ...init,
-    }),
-  );
-}
-
-/**
- * Record every `contextmenu` that reaches the listener, *as the
- * listener sees it*.
- *
- * `target` is retargeted for the scope reading it, so an assertion made
- * after dispatch has finished reports the shadow host however the event
- * was dispatched - which is the same answer a broken implementation
- * gives. It has to be read from inside the handler, where the component
- * reads it.
- */
-function recordMenus(el: EventTarget): { event: MouseEvent; target: EventTarget | null }[] {
-  const seen: { event: MouseEvent; target: EventTarget | null }[] = [];
-
-  el.addEventListener('contextmenu', (e) => {
-    e.preventDefault();
-    // Every real handler does this; the gesture must work anyway.
-    e.stopPropagation();
-    seen.push({ event: e as MouseEvent, target: e.target });
-  });
-
-  return seen;
-}
-
-describe('long-press opens a context menu', () => {
-  beforeEach(() => {
-    uninstall = installLongPressContextMenu();
-    ({ host, inner } = mountRow());
-  });
-
-  afterEach(() => {
-    uninstall?.();
-    uninstall = null;
-    host.remove();
-  });
-
-  it('dispatches one at the touch point, on the element touched', async () => {
-    const seen = recordMenus(inner);
-
-    press(inner, 'pointerdown');
-    await wait(HELD);
-
-    expect(seen).toHaveLength(1);
-    expect(seen[0]?.event.clientX).toBe(40);
-    expect(seen[0]?.event.clientY).toBe(60);
-    // Dispatched on the row itself, not on its shadow host - which is
-    // the difference between a per-row handler firing and only a
-    // delegated one firing.
-    expect(seen[0]?.target).toBe(inner);
-  });
-
-  it('is cancelled by a press that moves', async () => {
-    const seen = recordMenus(inner);
-
-    press(inner, 'pointerdown');
-    press(inner, 'pointermove', {
-      clientX: 40 + MOVE_TOLERANCE_PX + 5,
-      clientY: 60,
-    });
-    await wait(HELD);
-
-    expect(seen).toHaveLength(0);
-  });
-
-  it('tolerates the jitter a finger cannot help', async () => {
-    const seen = recordMenus(inner);
-
-    press(inner, 'pointerdown');
-    press(inner, 'pointermove', { clientX: 43, clientY: 62 });
-    await wait(HELD);
-
-    expect(seen).toHaveLength(1);
-  });
-
-  it('is cancelled by lifting early, and by a scroll', async () => {
-    const seen = recordMenus(inner);
-
-    press(inner, 'pointerdown');
-    await wait(BRIEF);
-    press(inner, 'pointerup');
-    await wait(HELD);
-
-    expect(seen).toHaveLength(0);
-
-    press(inner, 'pointerdown');
-    press(inner, 'pointercancel');
-    await wait(HELD);
-
-    expect(seen).toHaveLength(0);
-  });
-
-  it('ignores a mouse, which has a right button of its own', async () => {
-    const seen = recordMenus(inner);
-
-    press(inner, 'pointerdown', { pointerType: 'mouse' });
-    await wait(HELD);
-
-    expect(seen).toHaveLength(0);
-  });
-
-  it('swallows the click that ends the gesture, and only that one', async () => {
-    let clicks = 0;
-
-    inner.addEventListener('click', () => {
-      clicks += 1;
-    });
-
-    press(inner, 'pointerdown');
-    await wait(HELD);
-    press(inner, 'pointerup');
-    inner.click();
-
-    expect(clicks).toBe(0);
-
-    // The next tap is a tap: on a phone that is the user choosing an
-    // item in the menu that just opened, so eating it would make the
-    // gesture useless.
-    press(inner, 'pointerdown');
-    press(inner, 'pointerup');
-    inner.click();
-
-    expect(clicks).toBe(1);
-  });
-
-  it('stands down where the browser fires its own', async () => {
-    const seen = recordMenus(inner);
-
-    press(inner, 'pointerdown');
-    await wait(BRIEF);
-    // Chromium does this itself on touch; WebKit and the Android
-    // WebView vary, which is the whole reason both halves exist. A
-    // test cannot dispatch a *trusted* event, which is why the module
-    // tells its own apart by identity rather than by `isTrusted`.
-    inner.dispatchEvent(
-      new MouseEvent('contextmenu', {
-        bubbles: true,
-        composed: true,
-        cancelable: true,
-      }),
-    );
-    await wait(HELD);
-
-    // One menu: the browser's. Not two.
-    expect(seen).toHaveLength(1);
-  });
-});

frontend/test/components/queue-overlay-mode.test.ts

@@ -29,8 +29,61 @@ import { shadow } from '@test/support/render';
 
 const wrappers: HTMLElement[] = [];
 
+let restoreMedia: (() => void) | null = null;
+
+/**
+ * Answer the shell's phone query with `phone` until restored.
+ *
+ * Stubbed rather than emulated, for the reason `search-dialog.test.ts`
+ * gives: the runner's viewport is fixed at 1280x800, and the panel
+ * reads `matchMedia` in `connectedCallback` precisely so a test can
+ * answer it first.
+ */
+/**
+ * Answering the phone query is not enough on its own: what decides
+ * whether the scrim exists is a `change` listener, and a stub whose
+ * `addEventListener` is a no-op leaves that listener untested — the
+ * whole suite stays green with it deleted. So the stub records the
+ * listeners and hands back a way to fire them.
+ */
+function stubPhone(phone: boolean): (next: boolean) => void {
+  const real = window.matchMedia.bind(window);
+  const listeners = new Set<(e: MediaQueryListEvent) => void>();
+  let matches = phone;
+
+  window.matchMedia = ((q: string) =>
+    q.includes('max-width: 599px')
+      ? {
+          get matches() {
+            return matches;
+          },
+          media: q,
+          addEventListener(_: string, fn: (e: MediaQueryListEvent) => void) {
+            listeners.add(fn);
+          },
+          removeEventListener(_: string, fn: (e: MediaQueryListEvent) => void) {
+            listeners.delete(fn);
+          },
+        }
+      : real(q)) as typeof window.matchMedia;
+
+  restoreMedia = () => {
+    window.matchMedia = real;
+  };
+
+  return (next: boolean) => {
+    matches = next;
+
+    for (const fn of listeners) {
+      fn({ matches: next } as MediaQueryListEvent);
+    }
+  };
+}
+
 afterEach(() => {
   for (const w of wrappers.splice(0)) w.remove();
+  restoreMedia?.();
+  restoreMedia = null;
 });
 
 /**
@@ -157,6 +210,78 @@ describe('the queue panel decides whether it can be a column', () => {
     }
   });
 
+  /**
+   * #171 — the scrim is a dismissal target, so it exists only where it
+   * has pixels to be tapped.
+   *
+   * Below 600px `.panel-content` is `width: 100%`, so the scrim is
+   * entirely underneath an opaque panel: measured at 424x439, host,
+   * panel and scrim all 424x318. Drawing it there is a `cursor:
+   * pointer` click target nobody can reach, and the queue is a screen
+   * at that width anyway (#55) — back and the close button are its ways
+   * out. Existence rather than `display: none`, because a hidden scrim
+   * is still an element carrying the handler.
+   */
+  it('draws no scrim at phone width, where it would have no reachable pixels', async () => {
+    stubPhone(true);
+
+    const el = await panelIn(424);
+
+    expect(el.overlay).toBe(true);
+    expect(el.shadowRoot?.querySelector('.scrim')).toBeNull();
+
+    // The way out a thumb can hit is still there.
+    expect(
+      shadow(el, '[data-testid="queue-close"]')?.getAttribute('aria-label'),
+    ).toBe('Close queue');
+  });
+
+  /**
+   * The 600–899 band is where the panel is a 320px column of a wider
+   * content area, so the scrim has uncovered pixels and #24's
+   * tap-outside-to-close is real. Same width as the overlay tests
+   * above, with the phone query explicitly answered `false`, so this
+   * fails if the scrim is ever dropped for every overlay.
+   */
+  it('keeps the scrim above phone width, where it can be tapped', async () => {
+    stubPhone(false);
+
+    const el = await panelIn(700);
+
+    expect(el.overlay).toBe(true);
+
+    shadow<HTMLElement>(el, '.scrim')?.click();
+    await el.updateComplete;
+
+    expect(el.open).toBe(false);
+  });
+
+  /**
+   * The scrim's existence comes from `matchMedia` rather than a
+   * stylesheet, which only holds up if the query is *listened* to — a
+   * panel opened on a desktop and carried across the breakpoint (a
+   * resized window, an unfolded phone) has to lose its scrim without
+   * being reopened. Nothing else in this file fires `change`, so
+   * deleting the listener leaves the whole suite green.
+   */
+  it('drops the scrim when the viewport crosses the breakpoint', async () => {
+    const setPhone = stubPhone(false);
+
+    const el = await panelIn(700);
+
+    expect(el.shadowRoot?.querySelector('.scrim')).not.toBeNull();
+
+    setPhone(true);
+    await el.updateComplete;
+
+    expect(el.shadowRoot?.querySelector('.scrim')).toBeNull();
+
+    setPhone(false);
+    await el.updateComplete;
+
+    expect(el.shadowRoot?.querySelector('.scrim')).not.toBeNull();
+  });
+
   /**
    * Escape belongs to the overlay, not to the queue. An inline panel is
    * beside the content rather than over it, so there is nothing to

frontend/test/components/search-dialog.test.ts

@@ -99,6 +99,25 @@ describe('<search-trigger>', () => {
     }
   });
 
+  it('meets the touch floor it was shipped four pixels under', async () => {
+    stubPhone(true);
+
+    // #57 created this as the phone's replacement for the header search
+    // box, so it exists *only* where there is a thumb -- and it shipped
+    // at 40x40 under a comment calling that "the smallest a touch
+    // target should be", which was the app's own 44px floor (#56)
+    // restated short rather than a second opinion about it. #186.
+    const el = await fixture('search-trigger');
+    const button = shadow<HTMLButtonElement>(el, '[data-testid="search-trigger"]');
+
+    expect(button).not.toBeNull();
+
+    const box = button!.getBoundingClientRect();
+
+    expect(Math.round(box.width)).toBeGreaterThanOrEqual(44);
+    expect(Math.round(box.height)).toBeGreaterThanOrEqual(44);
+  });
+
   it('names what the button will search', async () => {
     stubPhone(true);
 

frontend/test/components/seek-bar-hit-area.test.ts

@@ -0,0 +1,211 @@
+/**
+ * The seek bar's painted track and the thing you can hit are allowed to
+ * differ, and a slider is the clearest case where they should.
+ *
+ * On `now-playing-view` — the screen that exists so a phone has
+ * somewhere to seek from — the slider measured 261x6 on the reference
+ * device (#187). Six pixels is the whole of the drag target on the
+ * app's primary seeking affordance, against a 44px floor the app set
+ * for itself in #56 and holds to in the queue panel.
+ *
+ * Two separate faults, and the first is why the second was not obvious.
+ *
+ * **The phone rule had never applied.** `seek-bar`'s stylesheet asked
+ * for a 12px track below 599px and then set 6px in a plain `wa-slider`
+ * rule *written after it*. A media query adds no specificity, so the
+ * plain rule won at every width — which is `index.css`'s documented
+ * rule ("the phone section is last on purpose") reproduced inside a
+ * component's own stylesheet. The source said 12 and the device said 6.
+ *
+ * **And 12px would still be under the floor**, so the target is built
+ * around the track rather than by thickening it: padding on the part
+ * that carries the gesture, with margins cancelling it so the row does
+ * not grow.
+ *
+ * This is asserted against the *parsed stylesheet*, on
+ * `hover-affordance.test.ts`'s precedent and with the same limitation
+ * stated rather than hidden: no tier here renders at a phone width with
+ * a real `wa-slider` laid out, so what can be checked is the shape the
+ * browser built from the css`` literal. The pixel measurements that
+ * chose these numbers were taken on the device and are recorded on
+ * #187 and in the stylesheet's own comment — a number measured on a
+ * phone is not a number CI can assert.
+ *
+ * Which is the regression worth catching anyway. Both failures are
+ * invisible on a desktop: hoisting the block back above the plain rule
+ * renders identically at every width CI runs at, and it is exactly what
+ * a tidy-up does.
+ */
+import { describe, expect, it } from 'vitest';
+
+import '@components/audio-player/seekbar/seek-bar';
+import { fixture } from '@test/support/render';
+
+/** The app's touch floor, from #56. */
+const TOUCH_FLOOR = 44;
+
+/** The width below which the phone's rules apply. */
+const PHONE_QUERY = /max-width:\s*599px/;
+
+type Rule = { text: string; condition: string | null };
+
+/**
+ * Every rule in the element's own adopted stylesheets, flattened **in
+ * order**, which is the whole point here: the fault being guarded is a
+ * rule sitting in the wrong place, not a rule being absent.
+ */
+function rulesOf(host: Element): Rule[] {
+  const sheets = host.shadowRoot?.adoptedStyleSheets ?? [];
+  const out: Rule[] = [];
+
+  for (const sheet of sheets) {
+    for (const rule of Array.from(sheet.cssRules)) {
+      if (rule instanceof CSSMediaRule) {
+        for (const inner of Array.from(rule.cssRules)) {
+          out.push({ text: inner.cssText, condition: rule.conditionText });
+        }
+
+        continue;
+      }
+
+      out.push({ text: rule.cssText, condition: null });
+    }
+  }
+
+  return out;
+}
+
+/**
+ * The two px numbers of a `*-block` declaration, as [start, end].
+ *
+ * A symmetric pair is **serialised back as one value** — `padding-block:
+ * 16px 16px` reads as `padding-block: 16px` — so a naive pair-reader
+ * fails on the shorthand rather than on the thing it is checking, and
+ * says the wrong thing about why. That is not hypothetical: it is what
+ * the symmetric-padding reversion did while this test was being
+ * proved.
+ */
+function blockPair(text: string, property: string): [number, number] | null {
+  const declaration = new RegExp(`${property}:\\s*([^;]+)`).exec(text)?.[1];
+
+  if (declaration === undefined) {
+    return null;
+  }
+
+  const values = [...declaration.matchAll(/(-?[\d.]+)px/g)].map((m) =>
+    Number(m[1]),
+  );
+
+  const [start, end] = values;
+
+  if (start === undefined) {
+    return null;
+  }
+
+  return [start, end ?? start];
+}
+
+describe("the seek bar's phone rules", () => {
+  it('are last, so they are not silently overridden', async () => {
+    const el = await fixture('seek-bar', {});
+    const rules = rulesOf(el);
+
+    // A sweep that read nothing passes vacuously — the same first
+    // assertion icon-language.test.ts makes, for the same reason.
+    expect(rules.length).toBeGreaterThan(0);
+
+    const declaresTrackSize = (r: Rule) => /--track-size:/.test(r.text);
+
+    const lastUnconditional = rules.findLastIndex(
+      (r) => r.condition === null && declaresTrackSize(r),
+    );
+    const phoneOverride = rules.findLastIndex(
+      (r) => r.condition !== null && PHONE_QUERY.test(r.condition)
+        && declaresTrackSize(r),
+    );
+
+    expect(lastUnconditional).toBeGreaterThanOrEqual(0);
+    expect(phoneOverride).toBeGreaterThanOrEqual(0);
+
+    // A media query adds no specificity. Written first, it loses.
+    expect(phoneOverride).toBeGreaterThan(lastUnconditional);
+  });
+
+  it('give the slider a pointer target of at least the touch floor', async () => {
+    const el = await fixture('seek-bar', {});
+    const rules = rulesOf(el);
+
+    const track = rules.find(
+      (r) => r.condition !== null && PHONE_QUERY.test(r.condition)
+        && /--track-size:/.test(r.text),
+    );
+    const target = rules.find(
+      (r) => r.condition !== null && PHONE_QUERY.test(r.condition)
+        && r.text.includes('::part(slider)'),
+    );
+
+    expect(track).toBeDefined();
+    expect(target).toBeDefined();
+
+    const trackSize = Number(
+      /--track-size:\s*(-?[\d.]+)px/.exec(track!.text)?.[1],
+    );
+    const padding = blockPair(target!.text, 'padding-block');
+
+    expect(padding).not.toBeNull();
+
+    // The padding is on ::part(slider) rather than on the host because
+    // that inner div is what carries the gesture: it has the listener
+    // and the touch-action, and it is exactly the host's size, so
+    // padding the host grows a box that does not take the press.
+    const hitArea = trackSize + padding![0] + padding![1];
+
+    expect(hitArea).toBeGreaterThanOrEqual(TOUCH_FLOOR);
+  });
+
+  it('do not grow the row they sit in', async () => {
+    const el = await fixture('seek-bar', {});
+
+    const target = rulesOf(el).find(
+      (r) => r.condition !== null && PHONE_QUERY.test(r.condition)
+        && r.text.includes('::part(slider)'),
+    );
+
+    expect(target).toBeDefined();
+
+    const padding = blockPair(target!.text, 'padding-block');
+    const margin = blockPair(target!.text, 'margin-block');
+
+    expect(padding).not.toBeNull();
+    expect(margin).not.toBeNull();
+
+    // now-playing-view's vertical budget is fixed and #51 measured
+    // every pixel of it: letting the row grow by the difference cost
+    // the album art 25px of 143 when it was tried on the device.
+    expect(margin![0]).toBe(-padding![0]);
+    expect(margin![1]).toBe(-padding![1]);
+  });
+
+  it('take the space above, because what is below is the transport', async () => {
+    const el = await fixture('seek-bar', {});
+
+    const target = rulesOf(el).find(
+      (r) => r.condition !== null && PHONE_QUERY.test(r.condition)
+        && r.text.includes('::part(slider)'),
+    );
+
+    expect(target).toBeDefined();
+
+    const pair = blockPair(target!.text, 'padding-block');
+
+    expect(pair).not.toBeNull();
+
+    const [above, below] = pair!;
+
+    // Measured at 424x439: the seek row is 19px and the play button's
+    // top edge is 8px below it, while `.art` above is a non-interactive
+    // div. A symmetric target would reach into the play button — the
+    // most important control on the screen — so the growth is upward.
+    expect(above).toBeGreaterThan(below);
+  });
+});

frontend/test/components/settings-touch-targets.test.ts

@@ -0,0 +1,425 @@
+/**
+ * Every control in Settings is at least 44px (#186, second pass).
+ *
+ * The header pass covered the five controls a user meets on every
+ * screen. Settings is the other half and is much the larger one: swept
+ * on the reference device (TLP301, 424x439) with all eleven
+ * `config-section`s expanded, **120 controls** were under the floor,
+ * not the 93 the issue's first table implies, and `config-field` — the
+ * row shape the issue names — is eight of them. The bulk is behind the
+ * disclosures:
+ *
+ * | control | size | count |
+ * |---|---|---|
+ * | `.column-arrow-btn` | **16x14** | 36 |
+ * | `.column-toggle` | 16x16 | 29 |
+ * | `shortcut-capture` button | 80x**25** | 26 |
+ * | download format checkbox | 16x16 | 8 |
+ * | `config-field` select | 335x**30** | 7 |
+ * | `wa-input` / `wa-button` | 204x**20**, 185x**21** | 6 |
+ * | `library-filter` select | 120x**32** | 1 |
+ * | `.overflow-btn` | 31x31 | 1 |
+ *
+ * **This tier can measure it, unlike #187's seek bar**, for the reason
+ * the header pass gives: the controls are real elements and the rules
+ * are min-sizes, so a real Chromium rendering a real component gives
+ * the actual answer at any width. And unlike the header there is no
+ * overflow fit on this page, so nothing here needs the negative-margin
+ * treatment — height is free and the two square controls can simply be
+ * square.
+ *
+ * **What the sweep cannot see is written down here as a test rather
+ * than as a comment**, because it is the trap this whole issue keeps
+ * setting. Two controls are invisible to a walk of `button, select,
+ * input`: `config-field`'s toggle, whose `<input>` is
+ * `opacity: 0; width: 0; height: 0` so the thing a finger hits is the
+ * `<label>` around it (34x19, smaller than anything in either of the
+ * issue's tables), and `shortcut-capture`'s reset button, which renders
+ * only for a shortcut somebody has already rebound. Both are asserted
+ * by name below.
+ */
+import { beforeEach, describe, expect, it } from 'vitest';
+
+import '@components/config-page/config-field';
+import '@components/config-page/config-page';
+import '@components/config-page/download-clients';
+import '@components/config-page/shortcut-capture';
+import '@components/library-filter/library-filter';
+
+import { flush, stub } from '@test/support/harness';
+import { fixture, shadow, shadowAll } from '@test/support/render';
+
+/** The app's touch floor, from #56. */
+const FLOOR = 44;
+
+/**
+ * Everything a finger can hit, through every shadow root under `root`.
+ *
+ * It descends rather than querying one root because Settings is a tree
+ * of components — `config-page` renders `config-section`s holding
+ * `config-field`s and `shortcut-capture`s — and the defect was
+ * distributed across all of them. This is the device sweep, run here.
+ */
+function controlsUnder(root: Document | ShadowRoot | Element): { name: string; el: HTMLElement }[] {
+  const SELECTOR = 'button, select, input, [role="button"], [role="tab"], [role="switch"]';
+  const found: { name: string; el: HTMLElement }[] = [];
+  const seen = new Set<Element>();
+
+  const walk = (node: ParentNode, depth: number): void => {
+    if (depth > 20) return;
+
+    for (const el of Array.from(node.querySelectorAll('*'))) {
+      if (el.matches(SELECTOR) && !seen.has(el)) {
+        seen.add(el);
+
+        const box = el.getBoundingClientRect();
+        const style = getComputedStyle(el);
+        const rendered =
+          (box.width > 0 || box.height > 0) &&
+          style.visibility !== 'hidden' &&
+          style.display !== 'none';
+
+        if (rendered) {
+          const host = (el.getRootNode() as ShadowRoot).host;
+
+          found.push({
+            name: `${host ? host.tagName.toLowerCase() : 'root'} ${
+              (typeof el.className === 'string' && el.className) || el.tagName.toLowerCase()
+            }`,
+            el: el as HTMLElement,
+          });
+        }
+      }
+
+      if (el.shadowRoot) walk(el.shadowRoot, depth + 1);
+    }
+  };
+
+  walk(root as ParentNode, 0);
+
+  return found;
+}
+
+/**
+ * What a finger actually hits for `el`.
+ *
+ * For everything in this app that is one element, that is the element.
+ * A native checkbox is the exception and is why this function exists:
+ * it cannot grow its hit area without growing its paint, and a 44px
+ * checkbox is not what anyone wants — so a checkbox that has a label
+ * is targeted *through* the label, which is the fix the column lists
+ * and the download formats both use.
+ *
+ * The fallback is the checkbox itself, deliberately: a checkbox with
+ * no label is a 16px target and this must still say so.
+ */
+function hitTarget(el: HTMLElement): HTMLElement {
+  const input = el as HTMLInputElement;
+
+  if (input.type !== 'checkbox' && input.type !== 'radio') return el;
+
+  const wrapping = el.closest('label');
+  const root = el.getRootNode() as ShadowRoot | Document;
+  const associated = input.id
+    ? root.querySelector<HTMLLabelElement>(`label[for="${CSS.escape(input.id)}"]`)
+    : null;
+
+  return wrapping ?? associated ?? el;
+}
+
+/** The ones that miss the floor, reported with the numbers. */
+function tooSmall(controls: { name: string; el: HTMLElement }[]): string[] {
+  return controls
+    .map(({ name, el }) => {
+      const box = hitTarget(el).getBoundingClientRect();
+
+      return { name, w: Math.round(box.width), h: Math.round(box.height) };
+    })
+    .filter((c) => c.w < FLOOR || c.h < FLOOR)
+    .map((c) => `${c.name} ${c.w}x${c.h}`);
+}
+
+/**
+ * Open every disclosure under `host`, so the sweep can see the page.
+ *
+ * A collapsed `config-section` renders its body with `hidden`, so its
+ * controls measure 0x0 — which is exactly why the issue's first table
+ * lists seven Settings controls and the real count is 120.
+ */
+async function expandEverySection(host: HTMLElement & { updateComplete: Promise<unknown> }) {
+  const sections = shadowAll<HTMLElement>(host, 'config-section');
+
+  expect(sections.length, 'the page renders disclosures to open').toBeGreaterThan(0);
+
+  for (const section of sections) {
+    section.shadowRoot
+      ?.querySelector<HTMLButtonElement>('button[aria-expanded="false"]')
+      ?.click();
+  }
+
+  await flush();
+  await host.updateComplete;
+
+  for (const section of sections) {
+    await (section as HTMLElement & { updateComplete?: Promise<unknown> }).updateComplete;
+  }
+
+  return sections;
+}
+
+/** A control's own box, named so a failure says which and how small. */
+function boxOf(el: Element | null | undefined): string {
+  if (!el) return 'missing';
+
+  const box = el.getBoundingClientRect();
+
+  return `${Math.round(box.width)}x${Math.round(box.height)}`;
+}
+
+function meetsFloor(el: Element | null | undefined): boolean {
+  if (!el) return false;
+
+  const box = el.getBoundingClientRect();
+
+  return Math.round(box.width) >= FLOOR && Math.round(box.height) >= FLOOR;
+}
+
+describe('a config field is the shape every Settings row uses', () => {
+  it.each(['text', 'number', 'select', 'directory', 'color'] as const)(
+    'a %s field meets the floor',
+    async (type) => {
+      const el = await fixture('config-field', {
+        schema: {
+          key: 'k',
+          label: 'Music folder',
+          type,
+          options: [{ value: 'dark', label: 'Dark' }],
+        },
+        value: type === 'color' ? '#ffd43b' : '',
+      });
+
+      const controls = controlsUnder(el.shadowRoot!);
+
+      // A sweep that found nothing passes vacuously.
+      expect(controls.length).toBeGreaterThan(0);
+      expect(tooSmall(controls)).toEqual([]);
+    },
+  );
+
+  it('grows the toggle, which no sweep of inputs can see', async () => {
+    // The `<input>` is opacity: 0; width: 0; height: 0, so the walk
+    // above skips it as a zero-sized node -- and the thing a finger
+    // hits is the styling <label> around it, which measured 34x19 on
+    // the device. It is absent from #186's tables for exactly that
+    // reason, and it is smaller than everything in them.
+    const el = await fixture('config-field', {
+      schema: { key: 'x', label: 'Scan on startup', type: 'toggle' },
+      value: true,
+    });
+
+    const target = shadow(el, '.toggle-switch');
+
+    expect(boxOf(target)).toBe('44x44');
+  });
+
+  it('keeps the toggle painted at its old size, in its old place', async () => {
+    // A 44px pill is not what a switch should look like. The box is
+    // 44px and the paint is not: the slider is a child centred in it,
+    // and negative inline margins hand the extra width back so the
+    // pill stays flush with the inputs in the rows above.
+    const el = await fixture('config-field', {
+      schema: { key: 'x', label: 'Scan on startup', type: 'toggle' },
+      value: true,
+    });
+
+    const target = shadow(el, '.toggle-switch') as HTMLElement;
+    const slider = shadow(el, '.toggle-slider');
+
+    expect(slider!.getBoundingClientRect().height).toBeLessThan(FLOOR);
+
+    const style = getComputedStyle(target);
+    const handedBack =
+      parseFloat(style.marginInlineStart) + parseFloat(style.marginInlineEnd);
+
+    expect(handedBack).toBeLessThan(0);
+  });
+});
+
+describe('the shortcut editor', () => {
+  it('meets the floor', async () => {
+    const el = await fixture('shortcut-capture', {
+      action: 'play.toggle',
+      label: 'Play/pause',
+      currentKey: 'Space',
+      defaultKey: 'Space',
+    });
+
+    expect(meetsFloor(shadow(el, 'button'))).toBe(true);
+  });
+
+  it('grows the reset button, which only a rebound shortcut renders', async () => {
+    // Not in either of #186's tables, and it cannot be: a sweep of a
+    // freshly-installed app never sees it. It appears the moment
+    // anybody uses the feature.
+    const el = await fixture('shortcut-capture', {
+      action: 'play.toggle',
+      label: 'Play/pause',
+      currentKey: 'K',
+      defaultKey: 'Space',
+    });
+
+    const reset = shadow(el, '.reset-btn');
+
+    expect(reset, 'a rebound shortcut renders a reset button').toBeTruthy();
+    expect(meetsFloor(reset)).toBe(true);
+  });
+});
+
+describe('the library filter', () => {
+  it('meets the floor in both of its placements', async () => {
+    // One component, two mount points since #57 -- the desktop top bar
+    // and Settings -> Libraries -- so it reaches the floor once.
+    const el = await fixture('library-filter');
+
+    const select = shadow(el, 'select');
+
+    expect(select, 'the filter renders a select').toBeTruthy();
+    expect(Math.round(select!.getBoundingClientRect().height)).toBeGreaterThanOrEqual(FLOOR);
+  });
+});
+
+describe('download clients', () => {
+  beforeEach(() => {
+    stub('download.Service.ListProviders', []);
+    stub('download.Service.ProviderKinds', []);
+    stub('config.Config.GetDownloadPreferences', {});
+  });
+
+  it('gives Web Awesome form controls the floor through the library API', async () => {
+    // wa-input's control is inside somebody else's shadow root, so the
+    // height comes from --wa-form-control-height rather than from a
+    // rule of ours reaching in. A custom property inherits through a
+    // shadow boundary, which is what makes a :host declaration reach
+    // it -- and what makes it measurable here.
+    const el = await fixture('download-clients');
+
+    await flush();
+    await expandEverySection(el);
+
+    expect(getComputedStyle(el).getPropertyValue('--wa-form-control-height').trim()).toBe(
+      '44px',
+    );
+
+    // And the outcome, not only the mechanism. A sweep of `input`
+    // reports a wa-input at 204x**42** even when this is right,
+    // because the inner input sits *inside* the control's own 1px
+    // border -- Web Awesome sizes it
+    // `calc(--wa-form-control-height - border-width * 2)`. What a
+    // finger hits is `part=base`, measured at 238x44 on the device.
+    //
+    // This reaches into another library's shadow root, which
+    // `name-dialog.ts` only permits where the failure is bounded. It
+    // is bounded the other way here: this is a test, so a renamed
+    // part fails loudly rather than silently passing, which is the
+    // direction that costs nobody a device session.
+    const input = shadowAll<HTMLElement>(el, 'wa-input').find(
+      (w) => w.getBoundingClientRect().height > 0,
+    );
+
+    expect(input, 'the add form renders a wa-input').toBeTruthy();
+
+    const base = input!.shadowRoot?.querySelector('[part~="base"]');
+
+    expect(base, 'wa-input still calls its control box "base"').toBeTruthy();
+    expect(Math.round(base!.getBoundingClientRect().height)).toBeGreaterThanOrEqual(FLOOR);
+  });
+
+  it('makes each allowed-format checkbox label a target', async () => {
+    const el = await fixture('download-clients');
+
+    await flush();
+    await el.updateComplete;
+
+    // Every section starts collapsed, and a collapsed body is `hidden`
+    // — so its controls measure 0x0 and a sweep of an unexpanded page
+    // reports them all as fine. That is how the issue's first table
+    // came to list seven Settings controls when there are 120.
+    await expandEverySection(el);
+
+    const options = shadowAll<HTMLElement>(el, '.format-option');
+
+    expect(options.length).toBeGreaterThan(0);
+
+    const short = options
+      .map((o) => ({ label: o.textContent?.trim(), h: Math.round(o.getBoundingClientRect().height) }))
+      .filter((o) => o.h < FLOOR);
+
+    expect(short).toEqual([]);
+  });
+});
+
+describe('the whole Settings page', () => {
+  beforeEach(() => {
+    for (const path of [
+      'library.Library.GetAllLibrariesWithTrackCounts',
+      'jobs.Service.GetJobs',
+      'download.Service.ListProviders',
+      'download.Service.ProviderKinds',
+    ]) {
+      stub(path, []);
+    }
+
+    stub('config.Config.GetShortcuts', {});
+    stub('config.Config.GetDownloadPreferences', {});
+    stub('config.Config.GetThemeAccentColor', '#ffd43b');
+    stub('config.Config.GetThemeBackgroundShade', 'dark');
+  });
+
+  it('has no control under the floor with every section expanded', async () => {
+    // The device sweep, run here: eleven collapsed sections is what
+    // made the first table look like seven controls. The density is
+    // behind the disclosures.
+    const el = await fixture('config-page');
+
+    await flush();
+    await el.updateComplete;
+    await expandEverySection(el);
+
+    const controls = controlsUnder(el.shadowRoot!);
+
+    expect(controls.length).toBeGreaterThan(0);
+    expect(tooSmall(controls)).toEqual([]);
+  });
+
+  it('makes a column row a target by naming it, not by growing the checkbox', async () => {
+    // A native checkbox cannot grow its hit area without growing its
+    // paint. The label is the target instead -- which is also the
+    // argument config-field already makes for its own labels, and it
+    // is behaviour rather than annotation: the column's name is now a
+    // click target for its checkbox.
+    const el = await fixture('config-page');
+
+    await flush();
+    await el.updateComplete;
+    await expandEverySection(el);
+
+    const labels = shadowAll<HTMLLabelElement>(el, 'label.column-label');
+
+    expect(labels.length).toBeGreaterThan(0);
+
+    for (const label of labels) {
+      const target = label.htmlFor
+        ? el.shadowRoot!.getElementById(label.htmlFor)
+        : null;
+
+      expect(
+        (target as HTMLInputElement | null)?.type,
+        `${label.textContent?.trim()} names its checkbox`,
+      ).toBe('checkbox');
+      expect(
+        Math.round(label.getBoundingClientRect().height),
+        `${label.textContent?.trim()} is a target`,
+      ).toBeGreaterThanOrEqual(FLOOR);
+    }
+  });
+});

frontend/test/components/touch-gestures.test.ts

@@ -0,0 +1,440 @@
+/**
+ * The touch gestures (plan 019, #63; long-press from plan 016 B2).
+ *
+ * These run in a real browser with real event dispatch, which is the
+ * only place the two things that make this hard are true: the synthetic
+ * event has to cross a shadow boundary to reach the listener a
+ * component actually bound, and the suppressors have to tell a trusted
+ * event from ours at document capture without eating the one they exist
+ * to deliver.
+ *
+ * The timings are real rather than faked, because the thing under test
+ * *is* a timing, and 600 ms twice is cheaper than a fake-timer harness
+ * that would also have to fake the pointer events.
+ */
+import { describe, expect, it, afterEach, beforeEach } from 'vitest';
+
+import {
+  installTouchGestures,
+  LONG_PRESS_MS,
+  MOVE_TOLERANCE_PX,
+} from '@utils/touch-gestures';
+
+/** A press that has certainly resolved, either way. */
+const HELD = LONG_PRESS_MS + 120;
+
+/** A press that has certainly not. */
+const BRIEF = Math.round(LONG_PRESS_MS / 4);
+
+const wait = (ms: number) => new Promise((r) => setTimeout(r, ms));
+
+let uninstall: (() => void) | null = null;
+let host: HTMLElement;
+let inner: HTMLElement;
+
+/** A row inside a shadow root, which is where every menu in this app
+ *  is bound — an element in the light DOM would pass a weaker test. */
+function mountRow(): { host: HTMLElement; inner: HTMLElement } {
+  const el = document.createElement('div');
+  const root = el.attachShadow({ mode: 'open' });
+  const row = document.createElement('div');
+
+  row.textContent = 'a track';
+  root.append(row);
+  document.body.append(el);
+
+  return { host: el, inner: row };
+}
+
+function press(
+  el: EventTarget,
+  type: string,
+  init: PointerEventInit = {},
+): void {
+  el.dispatchEvent(
+    new PointerEvent(type, {
+      bubbles: true,
+      composed: true,
+      cancelable: true,
+      pointerType: 'touch',
+      isPrimary: true,
+      clientX: 40,
+      clientY: 60,
+      ...init,
+    }),
+  );
+}
+
+/**
+ * Record every `contextmenu` that reaches the listener, *as the
+ * listener sees it*.
+ *
+ * `target` is retargeted for the scope reading it, so an assertion made
+ * after dispatch has finished reports the shadow host however the event
+ * was dispatched - which is the same answer a broken implementation
+ * gives. It has to be read from inside the handler, where the component
+ * reads it.
+ */
+function recordMenus(el: EventTarget): { event: MouseEvent; target: EventTarget | null }[] {
+  const seen: { event: MouseEvent; target: EventTarget | null }[] = [];
+
+  el.addEventListener('contextmenu', (e) => {
+    e.preventDefault();
+    // Every real handler does this; the gesture must work anyway.
+    e.stopPropagation();
+    seen.push({ event: e as MouseEvent, target: e.target });
+  });
+
+  return seen;
+}
+
+describe('an unclaimed long press is still a context menu', () => {
+  beforeEach(() => {
+    uninstall = installTouchGestures();
+    ({ host, inner } = mountRow());
+  });
+
+  afterEach(() => {
+    uninstall?.();
+    uninstall = null;
+    host.remove();
+  });
+
+  it('dispatches one at the touch point, on the element touched', async () => {
+    const seen = recordMenus(inner);
+
+    press(inner, 'pointerdown');
+    await wait(HELD);
+
+    expect(seen).toHaveLength(1);
+    expect(seen[0]?.event.clientX).toBe(40);
+    expect(seen[0]?.event.clientY).toBe(60);
+    // Dispatched on the row itself, not on its shadow host - which is
+    // the difference between a per-row handler firing and only a
+    // delegated one firing.
+    expect(seen[0]?.target).toBe(inner);
+  });
+
+  it('is cancelled by a press that moves', async () => {
+    const seen = recordMenus(inner);
+
+    press(inner, 'pointerdown');
+    press(inner, 'pointermove', {
+      clientX: 40 + MOVE_TOLERANCE_PX + 5,
+      clientY: 60,
+    });
+    await wait(HELD);
+
+    expect(seen).toHaveLength(0);
+  });
+
+  it('tolerates the jitter a finger cannot help', async () => {
+    const seen = recordMenus(inner);
+
+    press(inner, 'pointerdown');
+    press(inner, 'pointermove', { clientX: 43, clientY: 62 });
+    await wait(HELD);
+
+    expect(seen).toHaveLength(1);
+  });
+
+  it('is cancelled by lifting early, and by a scroll', async () => {
+    const seen = recordMenus(inner);
+
+    press(inner, 'pointerdown');
+    await wait(BRIEF);
+    press(inner, 'pointerup');
+    await wait(HELD);
+
+    expect(seen).toHaveLength(0);
+
+    press(inner, 'pointerdown');
+    press(inner, 'pointercancel');
+    await wait(HELD);
+
+    expect(seen).toHaveLength(0);
+  });
+
+  it('ignores a mouse, which has a right button of its own', async () => {
+    const seen = recordMenus(inner);
+
+    press(inner, 'pointerdown', { pointerType: 'mouse' });
+    await wait(HELD);
+
+    expect(seen).toHaveLength(0);
+  });
+
+  it('swallows the click that ends the gesture, and only that one', async () => {
+    let clicks = 0;
+
+    inner.addEventListener('click', () => {
+      clicks += 1;
+    });
+
+    press(inner, 'pointerdown');
+    await wait(HELD);
+    press(inner, 'pointerup');
+    inner.click();
+
+    expect(clicks).toBe(0);
+
+    // The next tap is a tap: on a phone that is the user choosing an
+    // item in the menu that just opened, so eating it would make the
+    // gesture useless.
+    press(inner, 'pointerdown');
+    press(inner, 'pointerup');
+    inner.click();
+
+    expect(clicks).toBe(1);
+  });
+
+  it('stands down where the browser fires its own', async () => {
+    const seen = recordMenus(inner);
+
+    press(inner, 'pointerdown');
+    await wait(BRIEF);
+    // Chromium does this itself on touch; WebKit and the Android
+    // WebView vary, which is the whole reason both halves exist. A
+    // test cannot dispatch a *trusted* event, which is why the module
+    // tells its own apart by identity rather than by `isTrusted`.
+    inner.dispatchEvent(
+      new MouseEvent('contextmenu', {
+        bubbles: true,
+        composed: true,
+        cancelable: true,
+      }),
+    );
+    await wait(HELD);
+
+    // One menu: the browser's. Not two.
+    expect(seen).toHaveLength(1);
+  });
+});
+
+/**
+ * What plan 019 adds on top, and the one property that protects
+ * everything downstream: a long press nobody claims is unchanged.
+ */
+describe('a gesture is announced before it is acted on', () => {
+  beforeEach(() => {
+    uninstall = installTouchGestures();
+    ({ host, inner } = mountRow());
+  });
+
+  afterEach(() => {
+    uninstall?.();
+    uninstall = null;
+    host.remove();
+  });
+
+  it('does not synthesise a menu when the long press is claimed', async () => {
+    // This is the whole reason #63 could reassign the hold without
+    // touching one of the fourteen context menus: the lists that want
+    // selection mode claim it, and nothing else changes.
+    const menus = recordMenus(inner);
+    const presses: unknown[] = [];
+
+    inner.addEventListener('yj-long-press', (e) => {
+      presses.push(e);
+      e.preventDefault();
+    });
+
+    press(inner, 'pointerdown');
+    await wait(HELD);
+
+    expect(presses).toHaveLength(1);
+    expect(menus).toHaveLength(0);
+  });
+
+  it('announces a tap on the element touched, at the touch point', async () => {
+    const taps: { target: EventTarget | null; x: number; y: number }[] = [];
+
+    inner.addEventListener('yj-tap', (e) => {
+      taps.push({ target: e.target, x: e.detail.x, y: e.detail.y });
+    });
+
+    press(inner, 'pointerdown');
+    await wait(BRIEF);
+    press(inner, 'pointerup');
+
+    expect(taps).toHaveLength(1);
+    // The row, not its shadow host -- the difference between a
+    // delegated handler firing and a per-row one never firing.
+    expect(taps[0]?.target).toBe(inner);
+    expect([taps[0]?.x, taps[0]?.y]).toEqual([40, 60]);
+  });
+
+  it('lets an unclaimed tap through as an ordinary click', async () => {
+    // Every button, link and checkbox in the app depends on this. Only
+    // a *claimed* tap has its click swallowed.
+    let clicks = 0;
+
+    inner.addEventListener('yj-tap', () => {
+      /* seen, not claimed */
+    });
+    inner.addEventListener('click', () => {
+      clicks += 1;
+    });
+
+    press(inner, 'pointerdown');
+    await wait(BRIEF);
+    press(inner, 'pointerup');
+    inner.click();
+
+    expect(clicks).toBe(1);
+  });
+
+  it('swallows the click behind a claimed tap', async () => {
+    // Or playing a track would also select it, and the row would end
+    // up in both states at once.
+    let clicks = 0;
+
+    inner.addEventListener('yj-tap', (e) => e.preventDefault());
+    inner.addEventListener('click', () => {
+      clicks += 1;
+    });
+
+    press(inner, 'pointerdown');
+    await wait(BRIEF);
+    press(inner, 'pointerup');
+    inner.click();
+
+    expect(clicks).toBe(0);
+  });
+
+  it('does not announce a tap for a press that became a long press', async () => {
+    // A hold is one gesture, not a hold and then a tap on release.
+    const taps: unknown[] = [];
+
+    inner.addEventListener('yj-tap', (e) => taps.push(e));
+
+    press(inner, 'pointerdown');
+    await wait(HELD);
+    press(inner, 'pointerup');
+
+    expect(taps).toHaveLength(0);
+  });
+
+  it('does not announce a tap for a press that drifted', async () => {
+    // A drifted press is a scroll, and the virtualizer's -- not a tap
+    // that happened to move. This is the one that would make a list
+    // unscrollable if it were wrong.
+    const taps: unknown[] = [];
+
+    inner.addEventListener('yj-tap', (e) => taps.push(e));
+
+    press(inner, 'pointerdown');
+    press(inner, 'pointermove', {
+      clientX: 40,
+      clientY: 60 + MOVE_TOLERANCE_PX + 20,
+    });
+    press(inner, 'pointerup');
+
+    expect(taps).toHaveLength(0);
+  });
+
+  it('ignores a mouse entirely, for both gestures', async () => {
+    // plan 019 decision 1: the predicate is the pointer, per event. A
+    // mouse on a touchscreen keeps click-selects / double-click-plays
+    // on the very same row, which no viewport width can express.
+    const seen: unknown[] = [];
+
+    inner.addEventListener('yj-tap', (e) => seen.push(e));
+    inner.addEventListener('yj-long-press', (e) => seen.push(e));
+
+    press(inner, 'pointerdown', { pointerType: 'mouse' });
+    await wait(BRIEF);
+    press(inner, 'pointerup', { pointerType: 'mouse' });
+    press(inner, 'pointerdown', { pointerType: 'mouse' });
+    await wait(HELD);
+
+    expect(seen).toHaveLength(0);
+  });
+});
+
+/**
+ * The browser's own long press is a trigger, not a competitor.
+ *
+ * This is a device-only defect made checkable here. `long-press.ts`
+ * stood down when a trusted `contextmenu` arrived mid-press, which was
+ * right while both paths ended in a context menu. Once a hold can mean
+ * *selection mode*, standing down means the gesture silently does the
+ * old thing — and Chrome 113's Android WebView does fire its own, so
+ * on the reference device `yj-long-press` was never announced at all
+ * while every test in this tier passed.
+ *
+ * A test cannot dispatch a *trusted* event, which is exactly why the
+ * module tells its own apart by identity rather than by `isTrusted`:
+ * an untrusted one dispatched from here takes the same path the
+ * browser's does.
+ */
+describe("the browser's own long press", () => {
+  beforeEach(() => {
+    uninstall = installTouchGestures();
+    ({ host, inner } = mountRow());
+  });
+
+  afterEach(() => {
+    uninstall?.();
+    uninstall = null;
+    host.remove();
+  });
+
+  /** Stand in for the browser recognising the hold itself. */
+  function browserContextMenu(el: EventTarget): MouseEvent {
+    const e = new MouseEvent('contextmenu', {
+      bubbles: true,
+      composed: true,
+      cancelable: true,
+    });
+
+    el.dispatchEvent(e);
+
+    return e;
+  }
+
+  it('announces the gesture rather than standing down', async () => {
+    const presses: unknown[] = [];
+
+    inner.addEventListener('yj-long-press', (e) => {
+      presses.push(e);
+      e.preventDefault();
+    });
+
+    press(inner, 'pointerdown');
+    await wait(BRIEF);
+    browserContextMenu(inner);
+    await wait(HELD);
+
+    expect(presses, 'the hold reached the component').toHaveLength(1);
+  });
+
+  it('suppresses its menu when a component claims the gesture', async () => {
+    const menus = recordMenus(inner);
+
+    inner.addEventListener('yj-long-press', (e) => e.preventDefault());
+
+    press(inner, 'pointerdown');
+    await wait(BRIEF);
+    browserContextMenu(inner);
+    await wait(HELD);
+
+    // The component wants selection mode, so the menu must not also
+    // open -- otherwise the device shows both at once.
+    expect(menus).toHaveLength(0);
+  });
+
+  it('still opens exactly one menu when nobody claims it', async () => {
+    // The old behaviour, reached by asking instead of assuming. This
+    // is what leaves the card grids, Explore and the playlist rows
+    // untouched by #63.
+    const menus = recordMenus(inner);
+
+    press(inner, 'pointerdown');
+    await wait(BRIEF);
+    browserContextMenu(inner);
+    await wait(HELD);
+
+    expect(menus).toHaveLength(1);
+  });
+});

frontend/test/components/touch-selection.test.ts

@@ -0,0 +1,279 @@
+/**
+ * What a finger does to a track list (plan 019, #63).
+ *
+ * The desktop semantics being diverged from are real and stay: a click
+ * selects, a double-click plays. A finger has no second button and no
+ * modifier keys, so the primary action has to be the primary gesture —
+ * and the inversion is decided **per event**, off `pointerType`, not
+ * off a viewport width or a platform flag (plan 019, decision 1).
+ *
+ * That is what these assert: the same row, in the same component, at
+ * the same width, answering a mouse one way and a finger the other.
+ *
+ * There is deliberately no double-tap. Measured on the reference
+ * device, playing a track is ~100ms end to end, and a double-tap
+ * discriminator has to hold every tap for the app's own
+ * `DOUBLE_CLICK_GRACE_MS` of 250 before it can act — 3.5x the primary
+ * interaction, to reach a menu a long press already reaches.
+ */
+import { afterEach, beforeEach, describe, expect, it, vi } from 'vitest';
+
+import '@components/track-list/track-list';
+import '@components/selection-bar/selection-bar';
+
+import { calls, flush, resetHarness, stub } from '@test/support/harness';
+import { fixture, shadow, shadowAll } from '@test/support/render';
+import { installTouchGestures, LONG_PRESS_MS } from '@utils/touch-gestures';
+
+const HELD = LONG_PRESS_MS + 120;
+const BRIEF = Math.round(LONG_PRESS_MS / 4);
+const wait = (ms: number) => new Promise((r) => setTimeout(r, ms));
+
+let uninstall: (() => void) | null = null;
+
+function track(n: number) {
+  return {
+    FilePath: `/music/track-${n}.mp3`,
+    TrackName: `Track ${n}`,
+    ArtistName: 'An Artist',
+    Album: 'An Album',
+    Duration: 100 + n,
+    ID: n,
+  };
+}
+
+const TRACKS = [track(1), track(2), track(3), track(4)];
+
+/** Dispatch a pointer event as a finger would produce it. */
+function press(el: EventTarget, type: string, init: PointerEventInit = {}) {
+  el.dispatchEvent(
+    new PointerEvent(type, {
+      bubbles: true,
+      composed: true,
+      cancelable: true,
+      pointerType: 'touch',
+      isPrimary: true,
+      clientX: 40,
+      clientY: 60,
+      ...init,
+    }),
+  );
+}
+
+/** A whole finger tap: down, a moment, up. */
+async function tap(el: EventTarget) {
+  press(el, 'pointerdown');
+  await wait(BRIEF);
+  press(el, 'pointerup');
+  await wait(0);
+}
+
+/** A finger held still until the gesture resolves. */
+async function hold(el: EventTarget) {
+  press(el, 'pointerdown');
+  await wait(HELD);
+  press(el, 'pointerup');
+  await wait(0);
+}
+
+async function mountList() {
+  const el = await fixture('track-list');
+
+  (el as unknown as { tracks: unknown[] }).tracks = TRACKS;
+  await flush();
+  await el.updateComplete;
+  await wait(60);
+  await el.updateComplete;
+
+  return el;
+}
+
+function rows(el: HTMLElement): HTMLElement[] {
+  return shadowAll<HTMLElement>(el, '.track-row');
+}
+
+describe('a finger on a track row', () => {
+  beforeEach(() => {
+    resetHarness();
+    stub('library.Library.GetTracks', TRACKS);
+    stub('library.Library.GetAllLibrariesWithTrackCounts', []);
+    stub('config.Config.GetShortcuts', {});
+    stub('queue.Queue.SetQueue', null);
+    stub('queue.Queue.AddTracksToQueue', null);
+    uninstall = installTouchGestures();
+  });
+
+  afterEach(() => {
+    uninstall?.();
+    uninstall = null;
+    vi.restoreAllMocks();
+  });
+
+  it('plays the row it taps, rather than selecting it', async () => {
+    const el = await mountList();
+    const row = rows(el)[1];
+
+    expect(row, 'the list rendered rows').toBeTruthy();
+
+    await tap(row!);
+    await flush();
+
+    const queued = calls('queue.Queue.SetQueue');
+
+    expect(queued.length, 'a tap plays').toBe(1);
+
+    // From that row, in the list as displayed -- not a queue of one
+    // that stops when the song ends.
+    expect(queued[0]?.args[1]).toBe(1);
+    expect((queued[0]?.args[0] as string[]).length).toBe(TRACKS.length);
+  });
+
+  it('leaves the same row to a mouse, which still selects', async () => {
+    // The inversion is per event, so one component answers both
+    // pointers at the same width. A viewport rule cannot say this.
+    const el = await mountList();
+    const row = rows(el)[1];
+
+    row!.dispatchEvent(
+      new MouseEvent('click', { bubbles: true, composed: true }),
+    );
+    await el.updateComplete;
+
+    expect(calls('queue.Queue.SetQueue').length, 'a click does not play').toBe(0);
+    expect(rows(el)[1]?.getAttribute('aria-selected')).toBe('true');
+  });
+
+  it('enters selection mode on a long press, with that row selected', async () => {
+    const el = await mountList();
+
+    await hold(rows(el)[2]!);
+    await el.updateComplete;
+
+    const bar = shadow(el, 'selection-bar');
+
+    expect(bar, 'the action bar appears').toBeTruthy();
+    expect((bar as unknown as { count: number }).count).toBe(1);
+    expect(rows(el)[2]?.getAttribute('aria-selected')).toBe('true');
+  });
+
+  it('does not open a context menu when it enters the mode', async () => {
+    // The gesture is claimed, so the layer must not fall through to
+    // the synthetic `contextmenu` that every other surface still gets.
+    const el = await mountList();
+
+    await hold(rows(el)[0]!);
+    await el.updateComplete;
+
+    const menu = shadow(el, 'menu-surface');
+
+    expect((menu as unknown as { active?: boolean } | null)?.active ?? false).toBe(
+      false,
+    );
+  });
+
+  it('toggles rows while the mode is on, instead of playing them', async () => {
+    const el = await mountList();
+
+    await hold(rows(el)[0]!);
+    await el.updateComplete;
+    await tap(rows(el)[2]!);
+    await el.updateComplete;
+
+    expect(calls('queue.Queue.SetQueue').length, 'no track was played').toBe(0);
+    expect(
+      (shadow(el, 'selection-bar') as unknown as { count: number }).count,
+    ).toBe(2);
+  });
+
+  it('leaves the mode when the last row is deselected', async () => {
+    // Android's own lists do this, and here it matters more than
+    // convention: the mode changes what a tap means, so a mode holding
+    // nothing is a list where tapping does nothing and the bar that
+    // would explain it is showing a count of zero.
+    const el = await mountList();
+
+    await hold(rows(el)[0]!);
+    await el.updateComplete;
+    await tap(rows(el)[0]!);
+    await el.updateComplete;
+
+    expect(shadow(el, 'selection-bar')).toBeFalsy();
+  });
+
+  it('keeps the favourite icon a favourite icon', async () => {
+    // It is inside a row whose tap now plays, and it has been a 44px
+    // target since #56 -- so without the "a control inside the row
+    // owns its own tap" rule, that target silently becomes a second
+    // play button.
+    const el = await mountList();
+    const fav = rows(el)[1]?.querySelector('.fav-icon');
+
+    expect(fav, 'a row renders a favourite control').toBeTruthy();
+
+    await tap(fav!);
+    await flush();
+
+    expect(calls('queue.Queue.SetQueue').length, 'tapping it does not play').toBe(
+      0,
+    );
+  });
+
+  it('does not play a row the finger scrolled from', async () => {
+    // The failure this exists for makes the list unusable rather than
+    // merely wrong: every flick to scroll would start a track.
+    const el = await mountList();
+    const row = rows(el)[1];
+
+    press(row!, 'pointerdown');
+    press(row!, 'pointermove', { clientX: 40, clientY: 200 });
+    press(row!, 'pointerup');
+    await flush();
+
+    expect(calls('queue.Queue.SetQueue').length).toBe(0);
+  });
+});
+
+describe('<selection-bar>', () => {
+  it('renders nothing with nothing selected', async () => {
+    const el = await fixture('selection-bar', { count: 0 });
+
+    expect(shadow(el, '.bar')).toBeFalsy();
+  });
+
+  it('announces the count, which changes under the finger', async () => {
+    const el = await fixture('selection-bar', { count: 3, actions: [] });
+    const live = shadow(el, '[aria-live="polite"]');
+
+    expect(live?.textContent?.trim()).toContain('3 tracks selected');
+  });
+
+  it('names one track in the singular', async () => {
+    const el = await fixture('selection-bar', { count: 1, actions: [] });
+
+    expect(shadow(el, '[aria-live="polite"]')?.textContent?.trim()).toContain(
+      '1 track selected',
+    );
+  });
+
+  it('keeps every control at the touch floor', async () => {
+    // #56 and #186. A bar a thumb uses, in the one mode that only a
+    // thumb can enter.
+    const el = await fixture('selection-bar', {
+      count: 2,
+      actions: [{ id: 'play', label: 'Play', icon: 'play' }],
+    });
+
+    const buttons = shadowAll<HTMLElement>(el, 'button');
+
+    expect(buttons.length).toBeGreaterThan(0);
+
+    for (const button of buttons) {
+      const box = button.getBoundingClientRect();
+
+      expect(
+        Math.min(Math.round(box.width), Math.round(box.height)),
+        button.getAttribute('aria-label') ?? '',
+      ).toBeGreaterThanOrEqual(44);
+    }
+  });
+});

frontend/test/components/touch-targets.test.ts

@@ -0,0 +1,162 @@
+/**
+ * Every control a finger meets is at least 44px (#186).
+ *
+ * #56 sized the playback transport for a thumb and named 44px; the
+ * queue header keeps it; nothing else was resized. So the controls a
+ * user meets on *every* screen — the sort control, its direction
+ * button, the page actions, the overflow trigger and the phone's search
+ * button — sat between a third and two thirds of the app's own floor.
+ * Measured on the reference device (TLP301, 424x439): `page-sort` 99x23,
+ * `page-sort-direction` **28x21**, `page-actions-more` 38x27,
+ * `search-trigger` 40x40.
+ *
+ * Unlike the seek bar's target (#187), this one can be measured here
+ * rather than inferred from the stylesheet. There the painted track had
+ * to stay thin, so the hit area was grown past its own box and only a
+ * phone-width layout of a third-party slider could show it. Here the
+ * control *is* the target, so a real Chromium rendering a real
+ * `page-header` gives the actual answer — and because it is a `min-size`
+ * rather than a media query, the answer is the same at every width,
+ * which is what makes it checkable in this tier at all.
+ *
+ * That is also why there is no phone branch to test: a 44px control on
+ * a desktop is merely large, and a second declaration of what a phone
+ * shows is a second thing to keep in step.
+ */
+import { describe, expect, it } from 'vitest';
+import type { PageAction, PageHeader } from '@components/page-header/page-header';
+
+import '@components/page-header/page-header';
+import { fixture, shadowAll } from '@test/support/render';
+
+/** The app's touch floor, from #56. */
+const FLOOR = 44;
+
+const SORTS = [
+  { id: 'name', label: 'Name' },
+  { id: 'tracks', label: 'Tracks' },
+];
+
+function actions(): PageAction[] {
+  return [
+    { id: 'import', label: 'Import', icon: 'file-import', priority: 0, onSelect: () => {} },
+    { id: 'new', label: 'New Playlist', icon: 'plus', priority: 2, onSelect: () => {} },
+  ];
+}
+
+/** Every visible control in the header's own shadow root. */
+function controlsOf(el: PageHeader): { name: string; el: HTMLElement }[] {
+  return shadowAll<HTMLElement>(el, 'button, select')
+    .filter((c) => !(c as HTMLButtonElement).hidden)
+    .map((c) => ({
+      name: c.dataset.testid ?? (c.className || c.tagName.toLowerCase()),
+      el: c,
+    }));
+}
+
+function tooSmall(controls: { name: string; el: HTMLElement }[]): string[] {
+  return controls
+    .map(({ name, el }) => {
+      const b = el.getBoundingClientRect();
+
+      return { name, w: Math.round(b.width), h: Math.round(b.height) };
+    })
+    .filter((c) => c.w < FLOOR || c.h < FLOOR)
+    .map((c) => `${c.name} ${c.w}x${c.h}`);
+}
+
+describe("the page header's controls", () => {
+  it('all meet the touch floor', async () => {
+    const el = await fixture<PageHeader>('page-header', {
+      heading: 'Playlists',
+      count: 50,
+      countNoun: 'playlist',
+      sortOptions: SORTS,
+      sortField: 'name',
+      sortDirection: 'asc',
+      actions: actions(),
+    });
+
+    const controls = controlsOf(el);
+
+    // A sweep that found no controls passes vacuously — the same first
+    // assertion icon-language.test.ts makes, for the same reason.
+    expect(controls.length).toBeGreaterThan(0);
+
+    // The two that were smallest, named so a regression says which.
+    expect(controls.map((c) => c.name)).toContain('page-sort-direction');
+    expect(controls.map((c) => c.name)).toContain('page-sort');
+
+    expect(tooSmall(controls)).toEqual([]);
+  });
+
+  it('grows the target without growing the box, so the overflow fit is untouched', async () => {
+    // The regression this exists for, and it was a real one: growing
+    // the two square controls to 44px *wide* added 22px to the header,
+    // which fit at every width Chromium was checked at and clipped the
+    // overflow trigger at 320x600 in WebKit -- the engine closest to
+    // what ships, and the one no machine here can run. #69's fit pass
+    // measures inline size, so a taller control is free and a wider one
+    // is not.
+    //
+    // Negative inline margins are what keep the box out of it: the
+    // padding makes the target, and the margin gives the space back.
+    const el = await fixture<PageHeader>('page-header', {
+      heading: 'Playlists',
+      sortOptions: SORTS,
+      sortField: 'name',
+      actions: actions(),
+    });
+
+    el.style.width = '320px';
+
+    for (let frame = 0; frame < 3; frame += 1) {
+      await new Promise((r) => requestAnimationFrame(r));
+      await el.updateComplete;
+    }
+
+    for (const selector of ['.sort-dir', '.more-button']) {
+      const control = shadowAll<HTMLElement>(el, selector).filter(
+        (c) => !(c as HTMLButtonElement).hidden,
+      )[0];
+
+      expect(control, selector).toBeTruthy();
+
+      const style = getComputedStyle(control!);
+      const added =
+        parseFloat(style.marginInlineStart) + parseFloat(style.marginInlineEnd);
+
+      expect(added, `${selector} gives its extra width back`).toBeLessThan(0);
+    }
+  });
+
+  it('includes the overflow trigger, which is the route to the rest', async () => {
+    // At 320px the fit pass collapses actions into the menu, so the
+    // trigger is rendered — and it is then the only way to reach them,
+    // which makes it the last control that should be hard to hit.
+    const el = await fixture<PageHeader>('page-header', {
+      heading: 'Playlists',
+      sortOptions: SORTS,
+      sortField: 'name',
+      actions: actions(),
+    });
+
+    el.style.width = '320px';
+
+    for (let frame = 0; frame < 3; frame += 1) {
+      await new Promise((r) => requestAnimationFrame(r));
+      await el.updateComplete;
+    }
+
+    const more = shadowAll<HTMLButtonElement>(el, '.more-button').filter(
+      (b) => !b.hidden,
+    );
+
+    expect(more.length).toBe(1);
+
+    const box = more[0]!.getBoundingClientRect();
+
+    expect(Math.round(box.width)).toBeGreaterThanOrEqual(FLOOR);
+    expect(Math.round(box.height)).toBeGreaterThanOrEqual(FLOOR);
+  });
+});

loghandler_android.go

@@ -0,0 +1,18 @@
+//go:build android
+
+package main
+
+import (
+	"log/slog"
+
+	"yellowjacket/backend/androidlog"
+)
+
+// newLogHandler builds the handler slog writes through.
+//
+// On Android stdout is /dev/null, so devslog here writes the app's
+// entire diagnosis into a hole -- see backend/androidlog. logcat is
+// the platform's sink and this is what reaches it.
+func newLogHandler(opts *slog.HandlerOptions) slog.Handler {
+	return androidlog.New(opts)
+}

loghandler_default.go

@@ -0,0 +1,20 @@
+//go:build !android
+
+package main
+
+import (
+	"log/slog"
+	"os"
+
+	"github.com/golang-cz/devslog"
+)
+
+// newLogHandler builds the handler slog writes through.
+//
+// Off Android that is devslog to stdout, as it has always been. The
+// selection is a build tag rather than a runtime check so that a
+// desktop binary links no cgo for a platform it will never run on --
+// backend/androidlog's write is -llog, which does not exist here.
+func newLogHandler(opts *slog.HandlerOptions) slog.Handler {
+	return devslog.NewHandler(os.Stdout, &devslog.Options{HandlerOptions: opts})
+}

main.go

@@ -8,7 +8,6 @@ import (
 	"strings"
 	"sync/atomic"
 
-	"github.com/golang-cz/devslog"
 	"github.com/wailsapp/wails/v3/pkg/application"
 	"github.com/wailsapp/wails/v3/pkg/events"
 
@@ -110,14 +109,23 @@ func main() {
 	// create sLogger
 	loglevel := resolveLogLevel(isDev)
 
-	sLogger := slog.New(devslog.NewHandler(os.Stdout, &devslog.Options{
-		HandlerOptions: &slog.HandlerOptions{
-			Level: loglevel,
-		},
-	}))
+	sLogger := slog.New(newLogHandler(&slog.HandlerOptions{Level: loglevel}))
 	slog.SetDefault(sLogger)
 	sLogger.Info("starting yellowjacket", "version", version, "commit", commit)
 
+	// Android has no /tmp and gives an app no TMPDIR, so anything in
+	// this process that spills to a temporary file is handed a path that
+	// does not exist -- see system.UseTempDir. It runs here rather than
+	// beside UseHomeOverride above because it has something to say when
+	// it fails and the logger does not exist up there; what matters is
+	// that it is before NewYellowJacketApp, which opens the database.
+	//
+	// A failure is not fatal: it leaves the platform's own answer in
+	// place, which is what every release before this one ran with.
+	if err := system.UseTempDir(application.Mobile.StoragePath()); err != nil {
+		sLogger.Error("could not set up a temp directory", "err", err.Error())
+	}
+
 	// Start profiling server (pprof + trace). In production builds this
 	// is a no-op — the compiler eliminates all profiling code.
 	stopProfiler := profiling.Start(sLogger)

scripts/android-emulator.sh

@@ -276,8 +276,14 @@ cmd_logs() {
 	# The app's own tags plus the two that report its death.  Chasing a
 	# raw logcat here is hopeless: the emulator emits thousands of lines
 	# a second, almost all of them WindowManager transitions.
+	#
+	# `yellowjacket` is where the Go side's slog goes (backend/androidlog,
+	# #160).  It is a fixed tag rather than "$PKG", which is the whole
+	# point of it being fixed: the debug build's id carries a ".dev"
+	# suffix, so a tag derived from the id would be filtered out on the
+	# one build anybody debugging this app is running.
 	"$ADB" logcat -v time \
-		WailsBridge:V "$PKG":V GoLog:V AndroidRuntime:E DEBUG:V libc:F ActivityManager:I '*:S'
+		yellowjacket:V WailsBridge:V "$PKG":V GoLog:V AndroidRuntime:E DEBUG:V libc:F ActivityManager:I '*:S'
 }
 
 # Forward the WebView's devtools socket, so the page can be asked things.