Bug#970247: lintian: debian-changelog-has-wrong-day-of-week false negative regression
On Sun, Sep 13, 2020 at 07:42:57PM +0200, Axel Beckert wrote:
> https://salsa.debian.org/lynx-team/lynx/-/merge_requests/1 by
> @janitor-bot made me aware of two dates in lynx's debian/changelog (as
> of the just uploaded 2.9.0dev.6-1) having a wrong week day. According to
> "ccal" these weekdays are indeed wrong. (See below.)
>
> But neither the current Lintian in unstable (2.94.0) nor the lintian
> version in Buster (2.15.0) emits that tag. What I see on Debian
> Unstable:
>
> $ lintian /var/cache/pbuilder/result/lynx_2.9.0dev.6-1_amd64.changes
> I: lynx source: patch-not-forwarded-upstream debian/patches/21_do_not_strip_-g.diff
> I: lynx source: patch-not-forwarded-upstream debian/patches/30_build_path_in_binary.diff
> I: lynx source: upstream-metadata-missing-repository
> $ egrep -v '^#|^$' ~/.lintianrc
> display-experimental = yes
> display-level = >=pedantic
> color = always
> suppress-tags=debian-news-entry-uses-asterisk
> $
>
> But some lintian version in between (around 2020-05-09) must have
> emitted it twice, otherwise lintian-brush/Lintian-Janitor wouldn't have
> found it and fixed it in the above mentioned merge request.
FWIW lintian-brush will find issues on its own once it runs, so in
this case it may have been an issue that it has always existed in
lintian.
Scheduling in the janitor happens based on the issues that
lintian identifies, but lintian-brush subsequently edits the files
on its own accord.
I'm working to introduce more cross-checking (verifying that
when lintian-brush claimed it fixed an issue, it was actually
an issue that lintian reported), which would allow us to bugs
like this as well as in the other direction (when lintian-brush
claims to fix things that aren't actually a problem).
Cheers,
Jelmer
Reply to: