* Allow invited users to fetch state events and events by ID
* Move failing test case to blacklist
* Fix /sync panicking due to presence udpates
* Use upload-artifact v3
* Yet another attempt to fix sytest artefacts
* Maybe this will work?
* Refactor ApplicationServiceWorkerState to be more robust
* Add launch.json to VS Code
* Implement login with JWT, registering with email, failed login rate limiting and reset password with m.login.email.identity auth type
* Log errors when JWT parsing failed
* Development build script
* Fix linter errors
* Use golangci-lint as a linter in VS Code
* Fix tests with RtFailedLogin
* Pass config load tests - parse JWT public key only if enabled
* Reduce CI steps
Do not support 386 arch and go 1.16, 1.17
* Fix linter errors
* Change RtFailedLogin logic - nil pointer can be provided
* Respect access token in query
* Fix typos
* Use only one mutex in RtFailedLogin
* Remove eventsRemaining across appservice component
* Push dendrite to production registry as well
* Rafactor TestRtFailedLogin
* Add database namespacing for unit tests
Background: Running `go test ./...` will run tests in different packages concurrently.
This can be stopped or limited by using `-p 1` (no concurrency). We want concurrency,
but this causes problems when running Postgres DBs in CI. The problem is that, in CI,
we have 1x postgres server exposing 1x postgres DB, which we wipe clean at the end of
each test via `defer close()`. When tests run concurrently, calls to `close()` will
delete data/tables which other tests are currently using, causing havoc.
Fix this by:
- Creating a database per package.
- Namespacing the database name by a hash of the current working directory (the directory containing those `_test.go` files)
This is exactly what SQLite does, quite unintentionally, via the use of `file:dendrite_test.db`,
which dumps the file into the current working directory which is the package running the tests,
hence deleting the file is safe when running concurrently.
* Linting
* Don't create the database in a txn
* dupe db is not an error
* syncapi: add more tests; fix more bugs
bugfixes:
- The postgres impl of TopologyTable.SelectEventIDsInRange did not use the provided txn
- The postgres impl of EventsTable.SelectEvents did not preserve the ordering of the input event IDs in the output events slice
- The sqlite impl of EventsTable.SelectEvents did not use a bulk `IN ($1)` query.
Added tests:
- `TestGetEventsInRangeWithTopologyToken`
- `TestOutputRoomEventsTable`
- `TestTopologyTable`
* -p 1 for now
* Add test infrastructure code for dendrite unit/integ tests
Start re-enabling some syncapi storage tests in the process.
* Linting
* Add postgres service to unit tests
* dendrite not syncv3
* Skip test which doesn't work
* Linting
* Add `jetstream.PrepareForTests`
Co-authored-by: Neil Alexander <neilalexander@users.noreply.github.com>
* Initial test
* Move CI to GHA
* Naming
* Always report all linter issues
* Remove if true
* Test complement in different variations
* Try again
* Move Complement back after initial tests and readd timeout
Make linting fail further checks
Remove CodeQL
* Update and rename tests.yml to dendrite.yml
Co-authored-by: Neil Alexander <neilalexander@users.noreply.github.com>
