Stop parsing urls for tryjobs on findit, instead provide an api on luci to get the appropriate information. |
||
Issue descriptionFindit relies on a set of hardcoded regexes to get the master, builder and buildnumber for builds and tryjobs. Instead of trusting this fragile hardcoded logic, it should instead call an appropriate api to get these parts from.
,
Nov 22 2017
1. From user input (less often, mostly for ourself to test), copy & paste a url to a failed build is convenient than selecting master/builder and paste a build number IMHO. 2. From SoM, but just the master url. 3. Also used to be from chromium-try-flakes, but CTF is not working now. So out of scope I think.
,
Nov 22 2017
1) looks like FindIt is willing to accept non-structure input for the sake of good UX. Then, IMO, that UI and only that UI should have the logic for parsing unstructured input. The rest of the FindIt should deal with structured data 2) sounds like SoM API should be improved, so that FindIt gets more structured response
,
Nov 22 2017
That makes sense, and thanks for the suggestions! just think of another one: 4) after a try-job started, buildbucket also returns a build url.
,
Nov 22 2017
yes, buildbucket returns build URL, but it also returns master, builder, and build number in "build_address" tag for LUCI builds. The format of build_address is "bucket/builder/build_number" or in future "project/bucket/builder/build_number"
,
Jan 23 2018
Chan: you are working on this already, right? |
||
►
Sign in to add a comment |
||
Comment 1 by no...@chromium.org
, Nov 22 2017