#303: depcheck: error: Incorrect number of arguments ----------------------+----------------------------------------------------- Reporter: kparal | Owner: tflink Type: defect | Status: assigned Priority: critical | Milestone: 0.4.7 Component: tests | Resolution: Keywords: | ----------------------+----------------------------------------------------- Comment (by jskladan):
Replying to [comment:6 tflink]:
I dug into this farther and the issue is a timing/coordination issue due
to how depcheck pulls in packages. Basically, depcheck doesn't limit itself to the nvrs passed in to the test - it pulls in everything from *-pending and tests what isn't already accepted.
The timeline (in UTC, all on 2011-04-11) is:
- 13:10:53 - depcheck-x86_64 (job 80736) scheduled for dist-f14
-updates-testing-pending
- Triggered by new builds for fuse-encfs-1.7.4-1.fc14 and
1:libguestfs-1.8.5-1.fc14
- 13:32:25 - sugar-paint-32-1.fc14 tagged dist-f14-updates-testing-
pending by bodhi
- 13:35:22 - job 80736 queued, depcheck pulls in all pending builds
(including sugar-paint)
- 13:37:15 - sugar-paint-32-1.fc14 passes depcheck, bodhi comment
created
- 13:40:50 - depcheck-x86_64 (job 80752) scheduled for dist-f14
-updates-testing-pending
- Triggered by build for sugar-paint-32-1.fc14
- 13:45:45 - job 80752 queued
- 13:46:19 - job 80742 fails because all builds for dist-f14-updates-
testing-pending are accepted
Great research! This was what I thought was happening (on the scheduling front), great to know that for sure.
Given this new information, the possible fixes I see are:
- Change depcheck.py so that it doesn't pull in any pending packages
that weren't specified
- Change depcheck so that it doesn't fail when no pending packages are
specified
I'd prefer the second possibility. If nothing else, it seems better to run depcheck just once, than to download the whole -pending tag just to test one package, which could have been already tested in the previous run.