Bug#954798: lintian: field-too-long checksums-sha256 error
Hi
On Tue, Mar 24, 2020 at 8:39 AM Chris Lamb <lamby@debian.org> wrote:
>
> > now Chris whitelisted this field from the check. But really the whole
> > file should be.
I have no issue disabling the check for buildinfo files. I just
thought that examining them was a stated goal, and your suggestions do
not work in that direction.
> Yeah, this just gets silly. :) Felix, is there something we are
> missing regarding why don't return to "sub binary" instead of "sub
> always" here?
I will disable this check for buildinfo files later today.
> (Regarding whether the check should even be here, I'm stepping out of
> that one I'm afraid but for reference it was added in #942493.)
buxy states in his report that *.changes files are less interesting
because those fields are copies. The same reasoning presumably applies
to buildinfo files. I am sure he won't mind. His bug was only about
Provides fields, anyway.
On a larger scale, the high number of complaints we received about
this check from the broader community (and from our own team members,
i.e. mattia) raise the question whether Lintian should be concerned
about output generated by dpkg and friends.
I agree with all of you (and have stated numerous times in the record)
that Lintian provides friendly packaging advice for the benefit of
maintainers. We are not dpkg's test suite. For that reason, I am happy
to remove this check from Lintian as long as an equivalent restriction
finds its way into dpkg.
Kind regards
Felix Lechner
Reply to: