Page MenuHomePhabricator

[Discussion] Disable whitelisted-user automated testing, unless requested.
Closed, DeclinedPublic


We currently test every patchset pushed by trusted users. This might not be the ideal behavior. This task proposes the hypothesis that we should follow the same rules as for non-whitelisted users' patches, roughly:

  • When pushing a new patchset, no tests are run regardless of who the author is.
  • When a trusted user CR+1 and/or leaves the gerrit comment "check all", "check lint", etc., run those tests.
  • When CR+2 a patchset, run gate-and-submit.

This will result in a dramatically lower number of tests being run, a problem reported in T225248, T224673, T204125. The question is whether the additional burden on developers to explicitly request tests (or CR+2 and merge pending test results) is worth the potential benefits of freeing up computational resources.

Event Timeline

awight created this task.Jun 17 2019, 1:52 PM
Restricted Application added a subscriber: Aklapper. · View Herald TranscriptJun 17 2019, 1:52 PM
hashar closed this task as Declined.Jun 19 2019, 7:20 AM
hashar added a subscriber: hashar.

I am being bold here and declining based on the proposal being a not so nice user experience. That adds way more manual steps when people have been used to get feedback automatically, and the tests will be run anyway, just that they will be run delayed. So that does not change much.