* Set `maxParallelForks = 1` for most tests, except for screenshot ones
It turns out the Robolectric setup penalty compounds the more parallel JVMs you spawn. Using 1 JVM vs 8 was up to 2.5x faster in local tests.
* Split the roborazzi tests tasks config and CI step
The problem seemed to be sometimes the background work to switch from live timeline to a focused event timeline wasn't finished in time and an assertion failed from time to time.
By providing the coroutine to run this on we can pass a `TestScope` and force it to reliably run immediately instead in unit tests.
The script was relying on the external git-flow tool and on a local
`git config gitflow.*` initialisation, which made its behaviour depend on
the machine setup and hid what was actually done to the branches.
The release branch is now created with `git checkout -b`, and the release
is finished with explicit merge, tag, back-merge and branch deletion
commands. The merge and tag messages are provided explicitly so that the
generated history stays identical to the previous releases. Each step now
aborts the script with a clear message when it fails, instead of letting
the script continue.
Note that tools/release/release.sh (V1) still uses git flow.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
* Try improving renovate issues with Maven Central rate limiting
Try not to make too many concurrent requests to Maven Central
The logs display rate limit responses followed by OK responses, which could mean we're hammering the Maven Central repo with too many concurrent requests
* Try not hitting Maven Central for `androidx` packages
sessionPaths keeps pointing at the directories of the account which has
just been logged in, and it was never reset. Adding another account
calls setHomeserver(), whose rotateSessionPath() starts by deleting the
directories of the previous sessionPaths, so the file and cache
directories of the first account were wiped.
finalizeClientCreation() now forgets the session paths once the login
succeeded. It is done there rather than in clear() on purpose: after a
failed attempt the directories are orphaned and must still be deleted by
the rotateSessionPath() of the next attempt.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Since the login flows stopped handing their client over to
RustMatrixClientFactory and let finalizeClientCreation() build a fresh
one from the SessionData, two problems appeared.
The client built by loginWithQrCode() was never owned by anyone: it is
not assigned to currentClient, so clear() had nothing to close and the
Rust Client stayed open for the whole app lifetime, holding the session
stores of the freshly logged in account. It is now tracked in
currentClient, and also disposed of when the QR login fails, which is
safe because that flow always builds its own client.
The client of the session was also built while the login client was
still open. Both use the same session paths, so their state and crypto
SQLite stores, and the search index when message search is enabled, were
opened twice at the same time with no cross process lock to protect
them. clear() now runs before the final client is created.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
* Update dependency org.matrix.rustcomponents:sdk-android to v26.09.08
* Fix API breaks:
- There's no `SyncServiceBuilder.withProfilesExtension()` method anymore. This is now enabled unconditionally in the SDK.
- Timeline filters no longer use the `org.matrix.rustcomponents.sdk.TimelineEventFilter` FFI wrapper, but the `uniffi.matrix_sdk_ui.TimelineEventFilter` type directly. This also changes the underlying types used for inclusion/exclusion.
---------
Co-authored-by: renovate[bot] <29139614+renovate[bot]@users.noreply.github.com>
Co-authored-by: Jorge Martín <jorgem@element.io>
Benoit Marty
ganfra
Jorge Martin Espinosa
renovate[bot]
Benoit Marty
ganfra