Internal: Backport AvoidCleanupTrait to 11.9.x #3469

We previously utilised a lot of clean-up steps to make sure our tests didn't affect eachother. However we've since moved to a scenario where both in the CI and locally a new database is loaded between scenarios, this makes the clean-up no longer needed. Additionally if the clean-up does run then this makes debugging a failed test a lot more difficult because you can't inspect the state of the platform at the time the test failed. This commit ensures that all the clean-up that is provided by the DrupalExtension by default is disabled. For this we have to disable autoloading of sub-contexts (which we weren't using anyway) because the search_api contained a `behat.inc` file that would load the `RawDrupalContext` class which contained more cleaning. There are DrupalExtension stories open to remove the cleaning from `RawDrupalContext`, but we don't have time to wait for that. We also add a step in our CI to dump the failed database to disk along with the installation.sql file. This allows a developer to load the failed database locally and see what was going wrong without having to rerun the test.

In newly written tests we were already doin this manually in each scenario. Between scenario's the user manager in the DrupalExtension was not correctly keeping track of the logged in user status which could prevent a new scenario from logging a user in. With the removal of clean-up in scenario's we now need that logout step literally everywhere, so it's easier to add it as an `@afterScenario` and run it every time than to require developers to decide about 100s of places whether it's needed. See jhedstrom/drupalextension#641

These were used in the past to help set-up test automation on Travis but they don't actually test Open Social. In any scenario that these tests might fail other tests would fail too. The tests are removed to reduce the number of parallel tests being executed.

This commit allows taxonomy reference data to be provided as a comma separated list of IDs or term names (mixing is allowed). This can now be used in tables for e.g. groups, topics and events. By putting this in the validate method we can use this everywhere. This ensures we deduplicate the taxonomy title lookup so that we don't break on a pre-fetched value. Since we now do it by field type we no longer need to do it by field name. We can also remove the topic clean-up since we load a new database before every test. This allows us to remove the helper that looks-up a topic type by name.

We can use our EntityTrait to allow any relative date for datetime fields that we create using any of our newer content scaffolding test steps.

These fields have a different field type and value structure than the `datetime` fields that we already tackled in a previous commit, so we need to add them separately.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Internal: Backport AvoidCleanupTrait to 11.9.x #3469

Internal: Backport AvoidCleanupTrait to 11.9.x #3469

Commits on Jul 22, 2023