Author: Sebastien Bacher
Author Date: 2023-01-24 08:58:19 UTC

build the new ruby3.1 on i386

Author: Brian Murray
Author Date: 2023-01-03 21:06:45 UTC

Don't comment on bugs if using no_act

Author: Steve Langasek
Author Date: 2023-01-01 22:29:00 UTC

small refactor, thanks Łukasz

Author: Brian Murray
Author Date: 2022-12-14 18:43:59 UTC

We also need SRVNAME in the cookie

Author: Steve Langasek
Author Date: 2022-12-07 00:09:38 UTC

Do not conflate RUNNING-REFERENCE state with a request to re-run a ref test

Currently when passing --state RUNNING-REFERENCE,
retry-autopkgtest-regressions will unconditionally construct a baseline test
for you (trigger=migration-reference/0). This prevents us using the tool to
construct a lists of tests that we want to retry /against the proposed
pocket/ from the list of those currently awaiting baseline retesting.

Previous behavior can still be achieved with --state RUNNING-REFERENCE
--no-proposed, but decoupling this also lets us do things like --state
RUNNING-REFERENCE --all-proposed with the expected behavior.

Author: Brian Murray
Author Date: 2022-11-30 20:33:10 UTC

Add in the canonical-ubuntu-qa team

Author: Brian Murray
Author Date: 2022-09-29 22:40:58 UTC

Readd budgie (oops!), remove some project names

Author: Brian Murray
Author Date: 2022-08-19 00:21:21 UTC

Use a set not a list

Author: Brian Murray
Author Date: 2022-07-25 18:08:17 UTC

Remove an extraneous line

Author: Steve Langasek
Author Date: 2022-07-16 04:26:56 UTC

Optimize the inner loop of phased-updater by not querying for extra binaries

Currently the code walks all of the binary publications for each SRU source,
but immediately afterwards there is a comment that "the p-u-p is currently
the same for all binary packages". This causes us to spend quite a lot of
time on API queries for other binaries, only to ignore the results.

Refactor the code to always only look at the first record returned from
getPublishedBinaries(), which has the same result *unless*
getPublishedBinaries() can return a mix of records with p-u-p and without;
and saves about 30% of the runtime of the script.

Author: Steve Langasek
Author Date: 2022-02-10 18:01:10 UTC

Fix nbs-report failure due to parsing error

The cron.NBS wrapper creates a file called 00FTBFS. If this file is
non-empty, it has a different format from all the other files in the target
directory. I don't know why this file is produced, but we shouldn't fail to
generate the report if the file is present.

Author: Steve Langasek
Author Date: 2022-01-10 22:57:56 UTC

PEP8 cleanup, thanks Łukasz!