Page MenuHomePhabricator

Show broken units in revision history

Authored by avivey on Sep 2 2016, 1:29 AM.
Referenced Files
Unknown Object (File)
Sat, Jun 25, 12:57 PM
Unknown Object (File)
Sat, Jun 25, 6:26 AM
Unknown Object (File)
Sun, Jun 12, 2:35 PM
Unknown Object (File)
Mon, Jun 6, 7:38 PM
Unknown Object (File)
Mon, Jun 6, 3:03 PM
Unknown Object (File)
Sat, Jun 4, 5:36 PM
Unknown Object (File)
May 23 2022, 2:10 AM
Unknown Object (File)
Apr 15 2017, 10:49 AM



This is hacky, and I'm not sure I'm happy with it; Until T9365 is done, this will show up
broken tests with an appropriate star in the Revision History.

Test Plan

Created 1M messages in a couple of old diffs in a revision. The query took ~80us (On SSD drive).

Diff Detail

rP Phabricator
Lint Not Applicable
Tests Not Applicable

Event Timeline

avivey retitled this revision from to Show broken units in revision history.
avivey updated this object.
avivey edited the test plan for this revision. (Show Details)
avivey edited edge metadata.
  • stray debug string
epriestley added a reviewer: epriestley.

Cool, I think this is pretty reasonable. Thanks!

(For context for anyone else reading this: this is a compromise to improve integration of Harbormaster unit test data in the short term, until we can fix this properly in D16417 or similar, but that requires more planning through T9365/T10635/T11402 than I want to do right away.)

It looks like there's no <buildTargetPHID, result> key on the harbormaster_buildunitmessage table so this might still need to look at a fair number of rows, but it shouldn't need to load them so I don't think it will make T10635 materially worse. We can keep an eye on it; I have more of an idea of how I want to attack that than of how I want to do aggregation.


(Is this a stray typo, or does it fix a display issue?)

This revision is now accepted and ready to land.Sep 2 2016, 12:02 PM
This revision was automatically updated to reflect the committed changes.