Pjotr Prins <pjotr.publi...@thebird.nl> skribis: > I asked this before. Would it be possible that hydra exports JSON (or > something) so we can get the package status without having to parse > HTML? So the output of > > http://hydra.gnu.org/job/gnu/master/ruby-2.2.3.x86_64-linux > > would be JSON with > > http://hydra.gnu.org/job/gnu/master/ruby-2.2.3.x86_64-linux.json
Everything is possible. :-) It turns out that Hydra has a bunch of things under lib/Hydra/Controller/API.pm¹. For example: http://hydra.gnu.org/api/nrqueue http://hydra.gnu.org/api/latestbuilds?nr=10 http://hydra.gnu.org/api/jobsets?project=gnu http://hydra.gnu.org/api/nrbuilds?nr=2&period=day&project=gnu&jobset=core-updates (Damn, we could even have an Emacs mode to display all this!) The /latestbuilds URL provides a “buildstatus” key, like this: { "project":"gnu", "timestamp":1442315120, "nixname":"gcr-3.16.0", "job":"gcr-3.16.0.mips64el-linux", "system":"mips64el-linux", "finished":1, "buildstatus":2, "jobset":"core-updates", "id":683368 } The meaning of the “buildstatus” value is given in hydra-postgresql.sql: -- Status codes: -- 0 = succeeded -- 1 = build of this derivation failed -- 2 = build of some dependency failed -- 3 = other failure (see errorMsg) -- 4 = build cancelled (removed from queue; never built) -- 5 = build not done because a dependency failed previously (obsolete) buildStatus integer, So it’s possible to query the build status of some of the latest builds. However, we’d need a different URL to get what you want. That seems to be quite simple to do. Would you or someone else be willing to hack on this? Another useful hack: in gnu.org/s/guix/packages, add a JS snippet that queries http://hydra.gnu.org/api/jobsets?project=gnu and displays some of the info. Thanks, Ludo’. ¹ https://github.com/NixOS/hydra/blob/master/src/lib/Hydra/Controller/API.pm
