* Make CI work on this internal fork
Every check was red on every pull request. Upstream declares permissions: {} in its
workflows, which costs nothing on a public repo and is fatal here: the token is
limited to Metadata:read, so actions/checkout cannot read an internal repo and dies
with a misleading "Repository not found". Repo settings cannot fix it — an explicit
permissions block always overrides the default.
Each job now gets the minimum it needs. Danger also reads the pull request through
the issues API, so it needs issues:read alongside pull-requests:write, and zizmor's
SARIF upload needs actions:read on a private repo.
The unit tests move to the org's Mac Studio: GitHub-hosted macos-26 could not finish
inside the hour and bills at 10x. Preview tests are skipped there, using the flag
upstream already uses for its own enterprise tests — they render a snapshot per
preview and never once finished on either runner. The trade is that preview snapshot
regressions are no longer caught in CI.
Workflows that only make sense upstream are gated off rather than deleted, so they
keep rebasing cleanly. The enterprise tests need Element's bot token and their
private submodule. Three nightlies fail at checkout every weeknight — this repo is a
detached copy rather than a network fork, so GitHub never disabled their schedules —
and one of them posts to Element's Codecov on every run.
* Erase the test simulators before running the tests
The unit tests assume a clean simulator, which costs nothing on a hosted VM
because it hands out a fresh device every run. This runner reuses the same two
devices, so app containers, caches and logs survive between runs and six suites
fail on the previous run's leftovers rather than on anything in the code.
LoggingTests is the clearest: it expects no log files and finds nine, in the same
device container each time.
Only the two devices the tooling pins are erased, resolved by UDID. A blanket
erase would take out simulator state belonging to other work on a machine the
release pipeline shares.
* Remove the simulator reset again
It was added on the theory that the runner's reused simulators were leaking state
between runs. That theory was wrong: erasing both devices changed nothing — the
same seven suites failed identically before and after.
The leftover files LoggingTests trips on are its own, from a previous attempt of
the same test within one run (the suite runs with -test-iterations 4), so nothing
crosses runs for an erase to clean. Keeping a step that mutates shared state on the
release machine for no measured benefit is worse than not having it.
* Run the tests on iOS 27.0
Bump the required simulator OS in the preview, UI and Compound tests, and the CI tool's default OS, from 26.5 to 27.0. Test devices stay the same.
iOS 27 writes transcoded videos 146 bytes larger, so update the expected sizes in MediaUploadingPreprocessorTests.
* Build and test on CI with Xcode 27
Select Xcode 27 in ci_common.sh and move every macOS workflow to the xcode-27 runner image, as macos-26 doesn't include Xcode 27. The Compound workflow now also creates its simulator on the iOS 27.0 runtime.
The xcode-27 runners are slower, so give the unit, Enterprise, Compound and snapshot recording jobs longer timeouts.
* Fix the preview snapshot layout on iOS 27
On iOS 27 the hosting controller applies the window's safe area when rendering, but the snapshot size comes from sizeThatFits, which ignores it. Component previews were shifted down and clipped at the bottom. Disable the hosting controller's safe area regions for these snapshots so the measured and rendered layouts match.
swift-snapshot-testing also renders into a window without a scene, and on iOS 27 a navigation stack in such a window reserves space for a large title it never shows, leaving a gap below the navigation bar. Attach the render window to the app's scene until swift-snapshot-testing handles it.
The invite users and add rooms screens sized their selected items row from a measured width stored in state, which is zero on the first layout pass, so snapshots could draw the row empty. Use containerRelativeFrame instead. The add rooms search previews now also wait for their results, and the search previews that show a clear button get a small precision tolerance, as it renders inconsistently in snapshots.
* Re-record the preview snapshots on iOS 27.0
Recorded on an iPhone SE (3rd generation) simulator running iOS 27.0 with Xcode 27.0 (27A266a). The changes come from iOS 27 rendering: text anti-aliasing, hyphenation and line breaking, navigation bars and toggles that now render in snapshots, lazy stacks that were blank on iOS 26, and pseudo-language locale formatting.
* Fix the UI tests on iOS 27
iOS 27 no longer exposes accessibility identifiers on SwiftUI menu items, so the space exploration test now finds them by label within the menu. On iPad, the start chat search field keeps focus after submitting and can swallow the next tap, so retry the tap if the create room screen doesn't appear.
* Re-record the UI test snapshots on iOS 27.0
Recorded on iPhone 17 and iPad (A16) simulators running iOS 27.0 with Xcode 27.0. The changes come from iOS 27 styling, like the keyboard height and glass borders. This also removes the iPad app lock setup snapshots, which have been orphaned since those tests stopped running on iPad.
* Fix the accessibility tests crashing on iOS 27
The accessibility tests' previews wrapper kept its ImageRenderer alive across the awaits that follow the render. On iOS 27, releasing it at that point tears down its view graph outside of a SwiftUI update, which hits an assertion and crashes the app. Release the renderer as soon as the image has been rendered.
* Keep the accessibility tests' image renderers alive on iOS 27
Releasing the ImageRenderer straight away wasn't enough: on iOS 27, releasing one that rendered a preview crashes SwiftUI whichever context it happens in. Keep the renderers alive instead. Each test launches a fresh app, so they only live for the one test.
* Share the UI tests' app settings with their mock user session
Since the user session flow reads its settings from UserSettings, the UI tests' mock user session was wrapping a fresh volatile AppSettings instead of the one the tests configure. The flags that skip onboarding weren't picked up, so every user session UI test got stuck on the notifications permission screen.
* Re-record the Compound snapshots on iOS 27.0
Apply the same safe area fix as the app's preview tests, so size-that-fits snapshots aren't clipped on iOS 27, and re-record the snapshots on an iPhone SE (3rd generation) simulator running iOS 27.0 in English (US).
* Fix a race reading the keyboard in the accessibility tests
On iOS 27 the QR code login screen's keyboard can be dismissing between checking that it exists and reading its frame, which fails the test. Snapshot it in a single query instead.
joel.poyet
renovate[bot]
Mauro
Mauro Romito
Stefan Ceriu