O8 still recommended adding xcodebuild -version to the release workflow as
future work. The Record toolchain step now does exactly that, so the risk row
described a gap that no longer exists and advised work already done.
The residual risk is narrower than O8 claimed: each run logs its own compiler,
but the log only reveals a runner upgrade after the fact rather than preventing
one.
unit-tests.yml moving to [self-hosted, macOS] on a pull_request trigger lets
any PR run arbitrary code on novaloop-mac-studio-001 — the same persistent
machine, keychain and signing identities the release pipeline uses. The risk
was accepted on the strength of the repo being INTERNAL, but that reasoning
lived nowhere: nothing in the repo said why it was tolerable, and nothing in
CI would ever say it existed.
Worth recording because the trade-off is invisible to the required checks.
zizmor reports self-hosted-runner only at --persona=auditor; the default
persona CI runs suppresses it, so a green build is not evidence of absence.
Verified both halves against the fix-fork-ci workflow: auditor reports it at
medium severity, the default reports "No findings to report … (3 suppressed)".
Records the exits too, so a future reader inheriting this need not rederive
them: restrict the trigger to non-fork branches, or return to GitHub-hosted
macos-26 with a raised timeout-minutes.
The plan embeds verbatim copies of two shipped files and both had drifted. The
shipped files are correct, so the copies are now byte-identical to them
(verified by cmp against `git show`).
Task 8's workflow still spliced `${{ github.event.inputs.version }}` straight
into the shell, so anyone re-deriving the workflow from the plan reintroduced a
script injection that the regex below it cannot prevent — the input is
interpolated before the check ever runs. It now reads the input via the
`INPUT_VERSION` env var, and gains the `Record toolchain` step that is a
release's only record of which compiler built it now that nothing pins Xcode.
Task 11's skill claimed deleting the app clears `hasSignedInBefore`. It does
not: the flag lives in the shared `group.ch.healthchat` UserDefaults suite,
which survives deletion, so that advice yields a void migration test that looks
like a pass — the precise false negative the section exists to prevent.
Task 3 Step 5 carried the same fault in a subtler form: it matched the bare
substring "skipping loadAccounts", but two lines in
`AuthenticationService.swift` (:52 and :57) end that way and mean opposite
things. Reading them as one told you to dismiss "Classic App not configured" —
migration broken for every user — as a void run. Step 1 already prescribed the
correct remedy and is unchanged.
Both documents told the reader to pin Xcode 26.4 and to install it via a
Brewfile. A probe of novaloop-mac-studio-001 refutes the premise: it carries
exactly one Xcode, 26.6, globally selected, with no versioned siblings and no
26.4 anywhere. The design justified the pin with "the runner has it and the
legacy pipeline proves it works" — both halves are false, and a pin naming an
absent version does not fail fast: xcodes starts a multi-GB authenticated
download that no CI secret can satisfy, so before_all hangs.
Record the no-pin decision and its cost — whatever Xcode the runner has is what
ships, unrecorded — rather than only deleting the wrong instruction, so nobody
reintroduces the pin from the same reasoning. Drop the Brewfile with it: its only
purpose was installing xcodes for that pin, and brew has no business running on a
shared release machine where it could upgrade xcodegen mid-release.
Add the runner's real constraints (no passwordless sudo, single Xcode, five
missing tools installed by hand) to the design, since they are why upstream's
setup_github_actions_environment cannot run here.
Two new risks. The legacy element-ios pipeline pins the same absent 26.4, so it
is very likely broken right now — not this fork's problem to fix, but the human
needs to know before the next legacy release. And with no pin, an Xcode upgrade
on the runner silently changes the compiler for every subsequent release.
The plan's embedded Fastfile and skill now match the branches that implement
them byte-for-byte, so the plan cannot drift from the artefacts it describes.
Task 6 Step 3 asserted the guard must fail before xcodes runs. It cannot:
fastlane's before_all always executes before the lane body, and the plan puts
xcodes in before_all and the guard in the lane. The acceptance criterion was
unmeetable on any machine.
The guard still does what it exists for — aborting before the ~30 minutes of
signing — so the wording was wrong, not the pipeline. Also record that the step
needs the xcodes CLI, since without it the lane dies earlier with an error that
reads like a config problem.
Task 11 described the skill's sections rather than showing them, the one place
the plan told an implementer what to write instead of what to type. Replace it
with the complete SKILL.md.
Writing it surfaced a broken convention worth not copying: the legacy skill
documents healthchat-v1.5.0-rc1 style tag suffixes, but the workflow derives the
marketing version by stripping only the healthchat-v prefix, so the suffix reaches
CFBundleShortVersionString, which Apple caps at three integers. Tags are numeric
only here, and test builds go through workflow_dispatch with the marker in the
release notes.
Task-by-task plan for the three PRs in the release design: the classic migration
fix, the fastlane pipeline, and the releasing-healthchat skill.
The migration fix gets a real test cycle — an existing upstream test,
classicAppAccountRequiresBackup, encodes the behaviour we're changing, so it is
rewritten first and drives the view model change.
Spec for bringing App Store release capability to element-x-ios, targeting the
existing HealthChat record that the legacy element-ios fork ships to today.
Both apps share bundle id ch.healthchat.element and team 6G37K72W82, so element-x
arrives as an in-place update rather than a new listing. That drives the design:
the identity already aligns and 26.07.3 clears the legacy's 1.4.1, but the repo
has no release machinery at all, and upstream's Xcode Cloud path is unavailable
because it needs the private Enterprise submodule.
Records three parts: repointing CLASSIC_APP_* at HealthChat's own identifiers so
existing users keep their sessions and encrypted history, a fork-owned fastlane
pipeline ported from the legacy fork, and a releasing-healthchat skill.
Revised after an adversarial review of every factual claim; 26 issues found, none
rejected. Most consequential: the migration fix is not monotonic, because making
loadAccounts() succeed exposes a backup-instructions sheet whose only button opens
element://open — a scheme this app does not register and whose legacy owner has
been replaced in place.
Applies the HealthChat product identity through the fork seams upstream already
provides, so no upstream logic files are touched and merges stay clean.
- Variants/HealthChat/healthchat.yml holds the identity scalars (display name,
bundle id ch.healthchat.element, app group, development team) and is included
permanently from project.yml. APP_NAME stays ElementX so the module name and
@testable import ElementX keep working.
- ElementX/Sources/AppHooks/HealthChat/ registers an AppSettings hook pinning
accountProviders to healthchat.ch with other providers disabled, push through
sygnal.healthchat.ch and a custom-scheme OIDC redirect; plus a Compound hook
recolouring the accent token family to the HealthChat blue.
- Entitlements are edited directly in ElementX/SupportingFiles/target.yml rather
than in the variant, because XcodeGen concatenates arrays on include-merge and
a variant can override scalars but cannot remove array entries. Drops the
Element Classic app group and keychain group, and narrows associated domains
to app.healthchat.ch, keeping matrix.to for permalinks.
- Info.plist CFBundleName now follows APP_DISPLAY_NAME — it, not
CFBundleDisplayName, is what the web authentication sign-in alert shows.
- App logo and app icon assets swapped in place; the onboarding welcome title
moves to a new screen_onboarding_welcome_title_ios untranslated string.
element-call-ios 0.1.0-rc.11 needs the SDK's 26.10.15 widget
capabilities, and brings matrix-rust-rtc 0.5.0-rc.5. The SDK drops
Room.roomEventsDebugString, whose mock goes with it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The call package now says why a call ended. Only our own hang-up is an
end call action; a call the other side declined or left, that nobody
answered, or that failed is reported to CallKit with that reason, so the
call log no longer shows it as one we ended, and no end action comes back
to hang up a call that is already over.
Also adopts the package's other host changes: the voice call icon beside
"Calling…", and the call's large avatar rendered at the 140 pt it is
drawn at.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Store the notifications switch per account
Adds AccountSettings.enableNotifications and moves the app-wide value into the signed in accounts before they're restored: the first account takes it out of the app settings and the others get the same switch, so someone who turned notifications off keeps them off.
* Add a notification service to every user session
Each account's NotificationService sets its pusher when its notifications are on and deletes it when they're off (appending with multiple accounts), keeps the account's unread count and decides which of its delivered notifications to clear. It asks the NotificationManager, the only owner of the notification center, through actions: remove the notifications matching a rule, request authorization and the unread count. The UserSessionManager connects every live session's service to the manager, and flows and screens use the session's service instead of the manager, which leaves the flow parameters.
* Route notifications by account
Taps only open rooms of the active account for now, a tap made while restoring only opens once its account is the active one, replies are sent from the account they were made for, and signing out only clears that account's notifications unless it was the last one. The notification extension hides pushes for an account that signed out while other accounts are still signed in.
* Turn notifications on and off for each account
The notification settings switch is now saved per account, and each account's NotificationService sets or deletes its pusher when it changes, asking for permission when it's turned on.
* Apply review suggestions to multi-account notifications
Migration
- Move the notifications switch with a single per-account migration, like 26.10.1 and 26.10.2, and drop its dedicated test.
Notification manager and service
- Follow each account's badge count publisher directly, removing the badge count action and the stored counts.
- Only reset the badge and unregister from APNs when the last account signs out, as each account already clears its own notifications.
- Set up the service's subscriptions in a single setupSubscriptions().
- Rename switchChanged(to:) to enableNotificationsChanged(to:) and pusherConfiguration(identifiers:) to pusherConfiguration(for:).
- Build the service without hopping to the main actor, it already runs there.
Notification service extension
- Check the account's switch in NotificationHandler using the session's settings, and log an error when a push arrives for an account that's turned off.
- Always show pushes that match no signed-in account and log an error, so a broken setup stays visible.
Other
- Store a pending inline reply in a PendingInlineReplyDetails struct.
- Assign the chats tab badge with weakAssign.
- Make PusherConfiguration Equatable so the pusher test checks it in one expectation.
- Clarify the comment about notifications without a receiver.
* Print instead of logging before the app hooks run in the NSE
A push that matches no signed-in account goes back to printing, as MXLog shouldn't be used before the app hooks have run. The push is still shown so a broken setup stays visible.
Tapping Continue and then the keyboard's Done key (or double tapping
Continue) started two server configurations. With a password-only server
both asked the flow to show the login screen, and the second request hit
the authentication state machine's unexpected transition fatalError.
Keep the running confirmation task and ignore further confirms until it
finishes, so the flow continues once. The guard clears afterwards so
confirming again after going back from the login screen still works.
Fixes ELEMENT-X-IOS-XED
XcodeGen (through XcodeProj) serializes schemes slightly differently to
Xcode, so Xcode rewrites them as soon as it touches them (running tests,
switching test plans, opening the scheme editor), leaving recurring local
changes behind.
Add a small Swift script, run from XcodeGen's postGenCommand, that rewrites
the generated schemes the way Xcode saves them: it drops default-valued
attributes and empty CommandLineArguments, orders TestPlanReference
attributes, and moves TestPlans to the end of the TestAction, omitting the
redundant empty Testables.
Regenerated schemes are byte-identical to the copies saved by Xcode. The
script can be removed once https://github.com/tuist/XcodeProj/pull/1189
ships in XcodeGen.
* Add a UserHooks system with just the CreateRoomScreenHook for now.
* Move more hooks into UserHooks.
* Pass the ClientProxy to makeUserHooks instead of calling AppHooks.configure(with:).
* Update the Enterprise submodule.
* Keep every signed-in account live
With the multi-account flag on, the accounts other than the active one are now
restored at launch, one after the other once the active account's UI is up.
They run their migrations, sync while the app is active, get a pusher (appended
so accounts on the same homeserver keep each other's) and count towards the app
badge, and background refresh and search backfill cover them too. Only the
active account's remote settings are ever applied.
Calls stay with the active account until the call service handles every account,
and an auth error on another account is only logged until accounts can be signed
out one at a time. With the flag off only the active account is live, as before.
* Sum the app badge across every signed in account
Each live account now reports its own unread count, and the app badge
shows their sum. Delivered notifications are only cleared for the
account whose rooms were read, so a room shared by two accounts keeps
the other account's notifications.
The room list observer for the badge and the notifications moves from
the UserSessionFlowCoordinator to the AppCoordinator's per-account
wiring. The flow keeps updating the chats tab badge.
* Let the user session manager manage every session
- The AppCoordinator tells the manager to pause, resume, go offline or
run the search backfill, instead of looping over the sessions itself.
- The manager also restores the other accounts, preparing each with the
AppCoordinator's migrations and wiring, and resumes them while the
services are running. With a single account nothing changes.
- Wiring an account again (e.g. after clearing the cache) replaces its
observers rather than adding to them.
- Add addAccountFlowEnabled, a developer-only switch for the upcoming
add account flow. It isn't a user preference, so testers enabling
multi-account can't add a second account yet.
* Apply review suggestions to keeping every account live
- Move the notification changes (pushers for every account, the summed
app badge and per-account clearing) to their own PR, so notifications
follow the active account again for now.
- Revert loading only the active account's remote settings: they load in
the store again, as on develop, until the hooks are reworked.
- The user session manager reports the auth errors of every live session,
so the AppCoordinator subscribes once instead of keeping per-account
observers.
- Restoring a single account is private to the manager now.
- Rename otherSessionsRestoreTask to otherSessionsRestorationTask.
* Restore every signed in account through one path
The user session manager now restores every account in a single pass,
most recently selected first, instead of separate restores for the
active account and the others. The AppCoordinator shows the first
account that comes back and no longer reads the multi-account flag:
without it, the manager only ever makes one account live.
- Clearing the cache clears every live account, and the next pass
replaces their sessions.
- Each account is migrated once per launch, and an account that just
signed in is never migrated.
- A pass also runs after signing in, so the other accounts come live.
- Remove UserSessionManagerError, restoreActiveSession and
restoreOtherSessions.
* Only clear the cache of the active account
Clearing the cache goes back to clearing the account shown in the UI.
Its session stays live until the next restore pass replaces it, and when
that restore fails another live account becomes the active one.
* Apply review suggestions to restoring every account
- Remove the search backfill guard: with no sessions the manager starts
nothing and reports that nothing is running, so the task completes.
- Add a TODO to switch to the most recently used account when the
active one signs out while others are signed in (element-meta#3325).
- Disable SwiftLint's type_body_length on the AppCoordinator again, it's
just over the limit with develop's new migrations.
* Move synchronous image upload preparation off the main actor
Signed-off-by: Ivan Magda <imagda15@gmail.com>
* Make media preparation cooperative with cancellation
Replace the dispatch queue with concurrent copy, image and video-thumbnail
functions. Return sendable upload metadata directly, bound each batch to
two preparations and discard unfinished results when cancellation occurs.
Keep file metadata reads synchronous and prevent the preview from sending
a result after its preparation task has been cancelled. Replace unrelated
tests with cancellation and batch-order coverage requested during review.
* Update MediaUploadingPreprocessor.swift
* Run media preparation in a nonisolated preprocessor
Make MediaUploadingPreprocessor a nonisolated struct whose copy, image
processing and video thumbnail run in @concurrent instance methods,
instead of a main-actor type with static helpers and an extra setting
parameter.
Check cancellation with plain early returns between the slow steps,
name the batch limit, stop logging cancelled video exports as errors,
and restore develop's structure and doc comments where they were changed
without need.
Replace the tests that didn't reach the new code with tests for cleanup
on failure, on batch failure and on cancellation, and make the preview
test set up its blocking mock before the view model starts processing.
* Apply review suggestions to media preparation
Docs:
- Drop the doc comment on maximumConcurrentProcessingCount, the name says it.
- Move the cleanup note into the main description of both processMedia
functions, so the Returns lines match develop again.
Tests:
- Rename the test helper to copyResourceToTemporaryDirectory(_:).
---------
Signed-off-by: Ivan Magda <imagda15@gmail.com>
Co-authored-by: Ivan Magda <imagda15@gmail.com>
* Remove the now unused setting to disable in-app notifications.
* Move visible account settings into AccountSettings.
Have ignored enableNotifications for now as that will require reworking the NotificationManager.
* Allow different sounds to be selected for each account.
* Move defaultNotificationTone into AccountSettings.
* Avoid repeated space membership scans in room search
Signed-off-by: Ivan Magda <imagda15@gmail.com>
* Refresh room candidates when space membership changes
Keep known space room IDs in a set updated by the room publisher and by
successful additions. Filter both suggestions and search results through
that set, retaining exclusions while the space list resets and reloads.
Preserve the initial recent-room eligibility filter so existing children
do not consume the five suggestion slots. Cover live membership updates
and partial-save filtering in both candidate lists.
---------
Signed-off-by: Ivan Magda <imagda15@gmail.com>
The forward, share, add-to-space and search pickers signal reaching the
bottom of the list with an empty `roomCount..<roomCount` range. The
prefetch sink added in 9dd9fd917 guards on `!range.isEmpty` to skip the
initial `0..<0`, which also dropped the pickers' bottom signal, so they
never loaded past the first 100 rooms.
Only skip `0..<0` instead, which leaves the home screen prefetch as it
was, and cover the bottom, prefetch and top cases in
RoomSummaryProviderTests.
* Ask for the model and harness used in AI-assisted PRs
Add an AGENTS.md rule that agents name the model and harness in the PR
description, and a matching line under the LLM checkbox in the pull
request template so human contributors fill it in too.
* Only let agents fill the LLM line of the PR template
Agents leave the description template to the developer and only fill in
the LLM line with the model and harness used.
* Bump element-call-ios to 0.1.0-rc.10
The release moves the RTC core to matrix-rust-rtc's room-first API, so the
stack no longer needs starting or stopping: it subscribes to a room when a
call opens it. It also adds headphones and Bluetooth audio output icons,
mapped here to Compound's headphones icon.
* Bump compound-design-tokens to 11.3.1 and use the Bluetooth output icon
11.3.1 adds the `bluetoothOutput` icon, replacing the `headphonesSolid`
placeholder for the Bluetooth audio route in native calls.
Sign-in now creates and stores its session through the
UserSessionManagerProtocol rather than the UserSessionStore, so the
AppCoordinator no longer holds the store.
- The manager forwards clientSessionDelegate and
userSession(for:sessionDirectories:passphrase:) to its store. As with
restoring, the new session isn't registered until the AppCoordinator
adds it.
- AuthenticationService takes the manager, for both sign-in and soft
logout.
- The store mock's configuration moves to the manager mock.
Set the chats tab's badge count from the same value used for the app badge.
The tab bar already renders it through the .badge modifier but that has no
effect outside of a TabView or List, so the iPad tab rail now overlays a
small NotificationCountBadge instead.
The sync update action is sent from the room list service state listener,
which fires before the SDK has applied the new unread counts, leaving the
app badge one sync behind. Drive both the badge and the clearing of
delivered notifications for fully read rooms from the room list publisher
instead, moving the observer from the ChatsTabFlowCoordinator into the
UserSessionFlowCoordinator.
The NotificationManager now receives the badge count from the caller so the
same value can be reused elsewhere.
Extract the unread notification count capsule from the room list cell into
its own view so it can be reused elsewhere. It caps the displayed value at
99+ and comes in a medium size (used by the room list) and a small size.
* Remove the dynamic member lookup from UserSettings in favour of nested access.
* Remove forwarding methods from UserSettings.
* Update the Enterprise submodule.
The state machine routes `.createdUserSession` from `.softLogout` to
`.signedIn`, but the AppCoordinator transition handler had no case for
it, so signing back in from the soft logout screen hit the
`fatalError("Unknown transition")` default.
Set up the user session as a restored one (`isNewLogin: false`): the
soft logout login reuses the same device ID and crypto store, and the
onboarding flow still starts if anything is outstanding.
The "every 2 weeks on wednesday" schedule opened a window for the whole day,
so after the group PR was merged Renovate kept running and raised new PRs on
the same branch for versions that passed minimumReleaseAge later that day
(e.g. #6282 followed by #6286 and #6293).
Restrict every schedule to before 4:00 am UTC so each group produces a single
PR per cycle. The "4:00 am" form is required, later.js silently drops the time
constraint for "4am" in this position.
joel.poyet
Valere
Mauro
Mauro Romito
Stefan Ceriu
Doug
renovate[bot]
Danny
Doug
Ivan Magda
Element CI
Valere Fedronic