We have several hundred accessibility tests that output platform-specific results. Sometimes when refactoring we need to update many tests, this is somewhat similar to how Blink layout tests work.

We're using this script to run try jobs and then collect the differences:

  tools/accessibility/rebase_dump_accessibility_tree_test.py

The current iteration of the script uses "git cl try-results" and "cit logdog query", but it's probably still not that robust, it still manually assembles the logdog query.

Any ideas on how we could make this more robust, i.e. so that changes to infrastructure are less likely to break it?

Are there infra tests? Could we write a test that would fail if somebody tried to make infra changes that would stop this script from working?