One of my goals is to review all the talos tests for accuracy and usefulness. It came to my attention earlier this summer that the dirty tests (tspaint_places_generated_med, tspaint_places_generated_max) are outdated and providing little to no value.
After investigating these tests in more details it became obvious that we were running 8 jobs/push at 15 minutes/job and ending up with the same numbers as ts_paint. Looking at the profile in detail, we have somewhat realistic data in the profile: https://wiki.mozilla.org/Buildbot/Talos/Tests#ts_places_generated_med The problem is no elements in the UI are accessing it. I thought a simple profile update would help, but our Firefox engineers have optimized the startup path so much that we would need to do more than a startup test to make this useful. In bug 771142 (https://bugzilla.mozilla.org/show_bug.cgi?id=771142), there is discussion about what we could do and what value this brings. After looking into this and asking interested parties, I am proposing to turn off these tests next week. Please comment in the bug or on this thread if you feel these tests should continue as startup tests. _______________________________________________ dev-platform mailing list dev-platform@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-platform
