Quality-review polish on Task 8:
1. InviteUnavailableError gains a typed public readonly `reason` field
(the union 'not found' | 'revoked' | 'expired' | 'exhausted'). Task 11
route handler can switch on the discriminant to produce user-facing
copy without parsing the message string.
2. The original "aborts transaction if insertUser throws" test was
vacuous — the callback threw before any DB write, so SQLite ROLLBACK
never fired and the post-conditions were trivially true. Replaced
with two tests: one that explicitly validates the synchronous
short-circuit (no DB work happens at all), and a second that writes
a real users row inside the callback then throws AFTER the write,
proving the SQLite ROLLBACK actually reverts the in-callback write.
Quality-review polish: a one-line comment over the empty-updates guard
in reinstateInvite explains why removing it would re-leak a confusing
Drizzle error. A new test verifies that when Path A (revoked->active)
fails its post-state check, the original token is preserved by the
SQLite transaction rollback (not replaced by the would-be new token).
Inside patchInvite/revokeInvite txn bodies, all reads now go through
the tx proxy. Pre-Task-7 hygiene: locks in the consistent pattern that
reinstateInvite (Task 7) and redeemInvite (Task 8) will copy.
Createinvite test failures now pin to InviteValidationError, catching
regressions where the wrong error class would otherwise pass silently.
Both functions wrap their read-modify-write in a Drizzle better-sqlite3
db.transaction((tx) => ...) with in-txn re-fetch so concurrent admin
mutations are serialized by SQLite's writer lock.
- patchInvite: 404 on missing, 409 on revoked, 400 on maxUses < usedCount,
allows expiresAt to be moved into the past (effective soft-shut).
- revokeInvite: 404 on missing, 409 on already-revoked (explicit reject,
not silent no-op).
Also extracts foldUsername() to collapse the duplicated
(username, isDeleted) -> display string fold across resolveCreatorUsername,
listInvites, and listRedemptions (deferred refactor from Task 5 review).
Note: the plan's example used db.transaction(cb)() with an IIFE,
which is the raw better-sqlite3 signature. Drizzle's wrapper
returns the callback's return value directly, so we use the
(tx) => ... form consistent with the rest of the codebase
(userDeletion, federation, channels, etc.).
Tests: 36 invite-service tests pass (27 prior + 9 new).
Full server suite: 40 files / 311 tests pass.
Adds two query helpers to inviteService:
- listInvites(filter): single-query LEFT JOIN against users to surface
createdByUsername, with status filtered in TS via the canonical
inviteStatus() derivation. Avoids N+1 the spec calls out (§3.1).
'archived' = expired | exhausted | revoked. Sort: createdAt DESC.
- listRedemptions(inviteId): LEFT JOIN against users via userId to
expose currentUsername alongside the registrantUsername snapshot.
Three null-handling branches per spec §3.1: live (username),
tombstoned ('Deleted User', isDeleted=true), and hard-deleted
(userId null, currentUsername null, isDeleted false).
Sort: redeemedAt DESC.
Also fixes a mistitled DB-miss test in getInviteByToken: the original
'returns null when token not found' used a 24-char string that fails
the format regex *before* the DB lookup. Split into two tests covering
both the format-reject path and the well-formed-but-missing path.
40 files / 302 tests passing.
Storage panel's max-upload-size input now accepts any positive integer
(bounded only by JS safe-integer ceiling) and offers an MB/GB unit
toggle. Multipart limit relaxed to MAX_SAFE_INTEGER — actual cap is
enforced per-request from the DB setting, not at the framework layer.
Wire broadcastStreamWatch (LiveKit data-channel ping) to all three
explicit user-action sites in StreamTile: "Stop Watching" context-menu
item, "Watch Stream" context-menu item, and the handleWatch button
callback. Broadcast is intentionally confined to click sites only —
not wired in voiceStore actions — to prevent noisy stream_user_left
sounds on automatic teardown paths in useLiveKit.ts.
vitest+jsdom puts test code in a different realm than Node's TextEncoder,
so `toBeInstanceOf(Uint8Array)` rejects the encoder's return value even
though it is structurally a Uint8Array. `Object.prototype.toString.call`
checks the @@toStringTag tag, which is realm-safe.
Strip the desktop package description to "Backspace" so Windows Task
Manager shows the bare product name instead of the long tagline.
Unify the web meta description and PWA manifest description on a single
positioning line that names Discord and TeamSpeak as the comparison
targets — improves link-preview copy and SEO surface.
Voice & Video tab no longer fires getUserMedia on mount. Uses
navigator.permissions.query({name:'camera'}) for state detection,
which is passive (no LED activation). Preview only opens when the
user explicitly clicks the dormant tile or the prompt-state CTA.
In-call mode unchanged (attaches existing LK track, no extra LED).
Spec and plan updated to reflect the corrected design — the
"auto-start preview, no Test Camera toggle" rationale was wrong;
macOS holds the camera LED on for ~2s after release, so any
incidental getUserMedia call (probe, transient mount) flashes the
LED in the user's face even when they aren't on the Voice & Video
panel. Privacy-correct UX: never light the LED without an explicit
user action.
Two-mode preview: in-call attaches to the LiveKit local camera track
(no double-capture); pre-call uses getUserMedia. Modes transition
reactively on isCameraOn changes.
Empty labels fall back to 'Camera N'; duplicate labels are
disambiguated with ' (1)', ' (2)' suffixes by enumeration order.
'No cameras detected' subline appears when enumeration is empty.
LocalTrackPublished registers a one-shot onended on the camera track's
MediaStreamTrack. The handler:
- skips when consumeIntentionalCameraOff() flag is set (user-initiated)
- re-probes getUserMedia to distinguish NotAllowedError (permission
revoked) from NotFoundError (disconnected) from other errors
- tears down camera state via the unified path
Also reset _intentionalCameraOff in voiceActions if setCameraEnabled(false)
rejects, so a failed disable doesn't poison the next genuine unplug.
Uses room.switchActiveDevice('videoinput', id) to swap the underlying
MediaStreamTrack without re-publishing. Compares against the published
track's actual getSettings().deviceId so null→explicit-same-device is a
no-op. Rolls back on failure with a 'Could not switch camera' toast.
- Voice-bar and mobile camera buttons now use the canonical handler
(fixes mobile no-op and voice-bar wrong-preset bugs)
- Remove dead useLiveKit.toggleCamera
- Add _intentionalCameraOff flag with mark/consume helpers
Two bugs in handleScroll's loadMoreMessages flow surfaced after the
smooth-scroll race fix.
(1) isLoadingMore stuck across channels. setIsLoadingMore(true) → await
loadMoreMessages → setIsLoadingMore(false) was unguarded. If the user
switched channels during the await, the new channel inherited the flag (same
component instance, same useState slot) and rendered the pagination skeleton
even with no load in flight. Cleared only when the original await resolved
or the component remounted (e.g., navigating to Friends and back).
(2) Wrong-channel scroll restore. The post-await rAF set
container.scrollTop = container.scrollHeight - prevScrollHeight against the
new channel's container with the old channel's prevScrollHeight, yanking
the new channel to a wrong position.
Fix:
- try/finally around the await so setIsLoadingMore(false) always runs.
- currentChannelIdRef tracks the live channelId; capture requestChannelId at
load start and compare both before scheduling the rAF and inside the rAF
callback (the 16ms frame gap is enough for a switch).
- Belt-and-suspenders: setIsLoadingMore(false) in the channel-switch effect
covers the case where the await never resolves (network hang). Without it,
a stuck await would leave the new channel inheriting the flag indefinitely.
No request cancellation — out of scope; AbortController plumbing through
chatStore is a bigger refactor and the channelId guard already silently
drops stale results.
Spec updated. Smooth-scroll fix from the previous commit untouched.
Smooth scrolls toward the bottom (new-message arrival in Effect A and the
Jump-to-Present click) animate scrollTop over many frames. Each intermediate
handleScroll measurement saw a large distanceFromBottom and flipped
isAtBottomRef to false, closing the Effect B/C gates. Lazy media (avatars,
embeds, Spotify thumbs) finishing mid-animation grew scrollHeight while the
gate was closed, so the smooth scroll landed at its originally-computed
target — leaving the user above the new bottom by ~the height of what loaded.
Fix: typed smoothScrollIntentRef ('bottom' | 'message' | null) with an 800ms
deadline. handleScroll suppresses the at-bottom flip while intent is 'bottom'
and the user hasn't wheeled past the 5000px nearBottom threshold. Effect D
fires a final defensive instant pin via native scrollend (Chrome 114+,
Safari 18+) or a setTimeout(800) fallback. 'message' intent (jump-to-message
from search) does NOT suppress — the gate flips honestly so the user is left
at the targeted message.
Verified live on nova.ddns.net Orbit → general: Jump-to-Present
lands flush at bottom; new Spotify-link messages stay at bottom as embeds
arrive via WS. docs/systems/message-list.md updated.
Iterating-by-paint-over wasn't working. Going back to first
principles: the brand IS the badge — full Element 1 mass, mark with
its own internal padding, dark frame around it. That's what reads as
"Backspace" on every other surface (dock, taskbar, favicon, apple-
touch). The sidebar home tile should look the same.
The earlier "ugly grey" complaint about Element 1 in the sidebar
came down to one specific colour mismatch: the source SVG's
#1d1d1b (warm near-black) sat next to the sidebar's #1a1a23 (cool
near-black) and read as an off-grey rectangle. Fix: in the generator,
do a string-replace `#1d1d1b → #000000` on app-icon.svg before
rendering logo.png only — every other output (.icns, .ico, favicons,
PWA) keeps the original brand `#1d1d1b`. Pure black against the
sidebar's #1a1a23 reads as a deliberately darker tile, not a hue
mismatch.
Reverts the SpaceSidebar lavender-tile workaround: with the badge
filling the tile via object-cover, the button's overflow-hidden +
rounded-[20px → 13px] morph already animates the badge cleanly.
backgroundStyle for type === 'dm' returns undefined again; deps trim
back to what they were.
Determinism gate verified — only logo.png changed (other 21 outputs
byte-identical across two regen runs).
Translucent white over the cool-dark sidebar reads as muddy grey;
that's the same root cause as the earlier #1d1d1b warm-grey complaint.
Any unsaturated tint loses against the dark sidebar bg.
Switching the home tile to accent-lavender (RGB 196,181,253 — the
endpoint of the mark's gradient) breaks the grey because saturation
defeats the grey-on-grey muddiness. Lavender also (a) complements
the gradient mark sitting on top, (b) parallels Discord's brand-
coloured home button pattern, and (c) gives the sidebar a clear
semantic colour hierarchy: action = mint, home = lavender (brand),
spaces = user-defined.
Three states escalate by alpha (0.08 / 0.16 / 0.28) so corner-morph
and fill-brightness both signal interaction.
The home/DM tile previously set backgroundStyle to undefined — the
button had no fill, so the existing rounded-[20px] → rounded-[13px]
corner-radius transition had nothing to morph and produced zero
hover/active feedback. The regression became visible after the icon
rebrand: the old logo.png shipped a #1d1d1b rounded-square that
filled the tile and gave it a de-facto background; the new bare-mark
logo is centred with transparent padding, exposing the missing
button bg.
Apply the same pattern as the existing 'action' tile: a translucent
white surface that brightens through three states (rest/hover/active)
so both the corner morph and the fill change are visible. Active
state (rgba 0.10) is the strongest because it signals 'you are on
the home page right now', complementing the existing pill indicator
on the left edge.
The previous Element-1 (full badge) approach produced a visible
warm-grey #1d1d1b square inside the cool-dark #1a1a23 sidebar — the
two near-blacks differ enough in hue to read as a foreign rectangle
in the slot.
The right answer (third time lucky): the SpaceSidebar's 40×40
rounded-[20px] tile already IS the dark squircle frame. logo.png now
composites the bare gradient mark at 75% on a transparent canvas:
the sidebar's tile bg shows through, the 25% padding keeps the mark
off the squircle edges and clear of the active-state ring, and there
is no foreign-bg colour conflict.
Refactored writeMaskablePng → writeCenteredMarkPng to share one code
path between the maskable PWA (opaque #1d1d1b background, 60% mark)
and the in-app logo (transparent background, 75% mark). Both use the
same composition pipeline; only canvas / scale / bg differ.
Only logo.png regenerated; other 21 outputs byte-identical
(determinism gate verified twice).
The SpaceSidebar's home tile is a 40×40 rounded-[20px] squircle with
object-cover. The original spec assumed the sidebar provides "the
colour-circle frame," so logo.png shipped as the bare gradient B on
transparent. In practice the tile is a transparent slot — the bare
mark touched the squircle edges (no internal padding), conflicted
with the active-state ring, and read as poorly-integrated.
Switching to the full badge (Element 1 / app-icon.svg): the badge's
own dark squircle bg fills the tile, its internal padding around the
mark keeps the gradient off the edges, and the result reads as a
clean Discord-style "home space" tile alongside the user's other
spaces.
Only logo.png regenerated; remaining 21 outputs byte-identical
(determinism gate verified). Spec output matrix and generator README
updated to reflect the corrected source mapping.
Jannis Braun