With accessibility semantics active, opening and leaving channels
retained disposed composer overlay states through Flutter’s semantics
traversal map. Those states kept departed route and message trees alive.
This upgrades the pinned Flutter SDK from 3.41.7 to 3.47.7, which
includes [the upstream
fix](3c96406cc4).
Includes SDK compatibility and formatter changes, preserves CocoaPods
integration, and gives create-channel radio lists a Material ink surface
required by the newer framework. Widget regressions verify both the
framework portal and the production channel/composer lifecycle: departed
routes, composer elements, and portal states are collectible while their
application’s semantics owner remains alive.
Related issue: no matching open issue or PR found.
### Validation
- [Current-head CI
passed](https://github.com/block/buzz/actions/runs/39085320351): all
mobile test shards, checks, Android APK, iOS release build, and native
simulator tests.
- Framework regression fails on Flutter 3.41.7 with three retained
states and passes on 3.47.7.
- Signed native Debug iOS Simulator build and paired-community
navigation replay: after 18 channel visits and forced GC, retained
disposed portal keys fell from 18 to 0 and route instances from 19 to 1.
- Review follow-up: all 365 focused channel/overlay tests, analyzer, and
complete mobile pre-push checks passed.
- Full local CI passed before the test-only review follow-up, including
Rust and desktop validation, frontend builds, analyzer checks, 3,499
mobile tests (four skipped), and the separate unconfigured-push test.
The replay used iOS 26.5 Simulator with accessibility semantics enabled,
CPU profiling disabled, and Debug/JIT builds. The before app was a
preinstalled emoji candidate; the after app uses this branch. Object
retention and the focused regression establish the fix; absolute heap
differences are not a controlled release-memory measurement. Android
native execution was not tested.
### Create-channel compatibility
The radio groups retain their layout after switching their container to
Material.
| Before (3.41.7) | After (3.47.7) |
| --- | --- |
| <img
src="https://raw.githubusercontent.com/block/buzz/42525e1bd9d9f264d0142c34ec61dbb6937ef758/pr-8419--01-before.png"
width="280" alt="Original create-channel radio groups"> | <img
src="https://raw.githubusercontent.com/block/buzz/42525e1bd9d9f264d0142c34ec61dbb6937ef758/pr-8419--02-after.png"
width="280" alt="Updated create-channel radio groups"> |
---------
Signed-off-by: Tom Brow <tomb@block.xyz>
Ordinary mobile messages eagerly built a set of every community emoji
shortcode and segmented their entire body just to decide whether to use
enlarged emoji rendering. Stop at the first non-emoji grapheme and build
the shortcode set only when a complete shortcode is encountered.
Emoji-only sizing and shortcode matching stay unchanged.
### Validation
- Full `just ci` passed.
- 125 focused detector and message-rendering tests pass; targeted
analysis is clean. The palette-access regression fails on the original
implementation (10,320 unnecessary reads) and passes with the fix. It
also checks that consecutive shortcodes share one palette scan.
- **Full Buzz app, real community data, debug/JIT iOS simulator, not a
rendering fixture:** two paired actual-input captures measured 125–127
µs per original detector call versus under 1 µs with the fix. All 5,200
executions per version agreed across 260 observed prose calls and a
2,580-entry palette. Instrumentation repeated each detector call, so
these captures are excluded from frame comparisons.
- Separate clean-code scrolling runs do **not** establish a reliable
smoothness improvement: quiet original UI-callback p95 was 10.1–12.6 ms
versus 10.1–12.9 ms fixed. A busy fixed run still reached 90.8 ms.
Normal app services stayed enabled; only scrolling was exercised. Live
background refresh, debug mode, host scheduling, and profiler overhead
limit interpretation. No physical-device/release-AOT or end-to-end
interaction-latency claim.
Signed-off-by: Tom Brow <tomb@block.xyz>
## Summary
Refresh mobile conversations with left-aligned grouped bubbles without
tails, tighter consecutive-message spacing, distinct sender colors,
theme-aware incoming backgrounds, and compact overlapping reaction
trays. Attach reply counts beside reactions with up to three participant
avatars; keep outgoing mentions and links readable in both themes.
Coordinate a 500 ms composer-to-message transition with spring
contraction and an anchored avatar. Message ownership follows the
signing identity so profile reloads cannot recolor outgoing bubbles.
Center channel activities without timestamps or icons, keep names
tappable, and show two affected members plus a **See all** sheet. Hide
legacy purpose activities, match media corners to bubble groups, and add
iOS status-bar scroll-to-top. This is an independent mobile PR; no
desktop changes or color picker.
## Validation
- Full `just mobile-check` passed; `just mobile-test` passed 3,275 tests
with 4 skips, plus 3 unconfigured-gateway checks.
- Fixed message-widget lint warnings and a stale notification fixture
that blocked analysis. Made the video-controller disposal test wait for
lifecycle events instead of fixed delays. Mobile pre-push checks are
enabled.
- Physical iPhone previews were installed throughout development. The
latest bubble, reply-avatar, contrast, and spacing updates were
installed and launched on the physical iPhone. Final motion and
status-bar behavior still need device visual verification; Android has
automated coverage but no physical-device verification.
---------
Signed-off-by: kenny lopez <klopez4212@gmail.com>
Brings the Flutter app's emoji and thread surfaces up to desktop parity.
## Emoji
- **Full emoji-mart dataset** generated from the same `@emoji-mart/data`
set desktop uses (1,870 emoji, 8 categories), committed as an asset — so
shortcodes, names, and keywords are identical across clients. `just
mobile-emoji-data` regenerates it.
- **Rebuilt the tray**: search (a Dart port of desktop's tiered
`emojiSearch` ranking, extended to names and keywords), a
frequently-used section, and one continuous scroll with pinned section
headers. The category rail is a shortcut into that list, not a page
switcher, and spans the full width the search field uses. Custom emoji
share the native glyph size and cell.
- **Reaction pills** match desktop's geometry, and the count shows at 1.
- **Emoji-only messages** render at 36px with 1.45em inline custom
emoji, matching desktop's `emojiOnly` treatment.
- **Positive-emoji burst** ported from desktop's `EmojiBurstProvider`,
suppressed under reduced motion.
## Threads
- **Top-down layout** — head first under the app bar, replies flowing
down, like desktop's thread panel. The old reversed list bottom-anchored
the content and jammed the head against the composer.
- **Tap a channel message to open its thread**; long-press still opens
the action sheet.
- **Live reactions.** The thread's relay query is one-shot and its
`kinds` filter carries only content rows, so a reaction event could not
reach an open thread at all, and `allMessages` was a snapshot frozen
when the route was pushed — a new pill only appeared after leaving and
re-entering, which refetched. The live channel events are now unioned
into the thread's list. The burst is also route-guarded, since the
channel timeline stays mounted underneath and was claiming it first.
- The `+` affordance follows the channel: replies stay bare until they
carry a reaction, and the head keeps a standing `+`.
## Keyboard
A deliberate downward drag past ~48px dismisses the keyboard; short
scrolls leave it alone. Applies to the channel list, the thread list,
and the compose bar (via a raw `Listener`, so it can't steal the field's
tap or selection drags).
True finger-tracking dismissal is out of scope — Flutter only offers
`manual`/`onDrag`, and 1:1 tracking needs a native `UIScrollView` proxy
plus Android's `WindowInsetsAnimationController`.
## Testing
`just mobile-check` and `just mobile-test` pass (965 tests). New
coverage for emoji search ranking, dataset parsing, the emoji-only
predicate, tray scroll/rail behavior, reaction pills and the burst, and
both thread fixes above. `just ci` green.
---------
Signed-off-by: kenny lopez <klopez4212@gmail.com>
Signed-off-by: npub12zsjdqx8dud99s9h47xmk9lq93vryf7zjrae8wdrmma52cg5yglseyulst <50a12680c76f1a52c0b7af8dbb17e02c583227c290fb93b9a3defb456114223f@buzz.block.builderlab.xyz>
Signed-off-by: klopez4212 <klopez4212@gmail.com>
Co-authored-by: npub12zsjdqx8dud99s9h47xmk9lq93vryf7zjrae8wdrmma52cg5yglseyulst <50a12680c76f1a52c0b7af8dbb17e02c583227c290fb93b9a3defb456114223f@buzz.block.builderlab.xyz>
