GitHub bug watches fail due to "id" vs. "number" confusion

[ԜаӀtеr Ⅼарсһуnѕkі]

This bug on Calamares:
https://bugs.launchpad.net/ubuntu/+source/calamares/+bug/1801440
includes a link to the upstream bug:
https://github.com/calamares/calamares/issues/712
which Launchpad is having problems importing with an unknown error:
Unknown error (Launchpad couldn't import bug)

However, the same bug tracker has no problems elsewhere:
https://bugs.launchpad.net/ubuntu/+source/calamares/+bug/1820960
https://bugs.launchpad.net/ubuntu/+source/calamares/+bug/1801498

[Colin Watson (cjwatson)]

This is e.g. OOPS-c8dd0400fd03ea651091bdf639a870f7, which looks like some kind of logic error in the GitHub import logic:

Traceback (most recent call last):
  Module lp.bugs.scripts.checkwatches.remotebugupdater, line 140, in updateRemoteBug
    self.remote_bug))
  Module lp.bugs.externalbugtracker.github, line 198, in getRemoteStatus
    remote_bug = self.bugs[int(bug_id)]
KeyError: 712

[Colin Watson (cjwatson)]

Hmm. I think the problem is that we're using the "id" field of the issue object returned by GitHub when we should be using the "number" field. I'm slightly confused about how this ever worked (which I'm sure it did at some point) - did GitHub change something or did we get lucky somehow?

[Launchpad QA Bot (lpqabot)]

Fixed in stable r18927 <http://bazaar.launchpad.net/~launchpad-pqm/launchpad/stable/revision/18927>.

[ԜаӀtеr Ⅼарсһуnѕkі (wxl)]

Will the affected bug reports self heal their watches? I've seen no change so far.

[Colin Watson (cjwatson)]

I haven't QAed or deployed the fix yet, so that's not surprising.

[Colin Watson (cjwatson)]

Deployed now. It may take a while (could be as much as a week, I think, since the bug watch scheduler backs off on failures), but the watches should self-heal. I've reset the ones mentioned in this bug report and in its duplicate.