don't forget, the tag could also be marked as experimental, which is
thought exactly to help out developing of the tags with high chances of
fpos.

On Tue, Feb 6, 2018 at 4:48 AM Daniel Kahn Gillmor <d...@debian.org> wrote:

> On Mon 2018-02-05 04:38:14 +0530, Chris Lamb wrote:
> > tags 889592 + pending
> > thanks
> >
> > "Fixed" in Git, pending upload:
> >
> >
> https://anonscm.debian.org/git/lintian/lintian.git/commit/?id=293c897ef968e0f50ac4f48986034aeda57e179d
> >
> > Ah well, just too many false-positive cases.. I mean, we'd have to start
> > ignoring ": " comments, as well as detecting the differences between,
> say:
> >
> >     override_dh_auto_test:
> >               echo "input" | ./testsuite.sh
> >
> > and
> >
> >     override_dh_auto_test:
> >               echo "Not running"
>
> Hm, ripping it out entirely seems suboptimal, i'd love to be able to
> find and fix a bunch of these.
>
> What if the tag didn't trigger if there was only one command in
> override_dh_auto_test, and it started with either "dh_auto_test " or
> with ": "?
>
> i think that would avoid most of the important false-positives, and the
> tag could provide guidance for folks who were doing 'echo "Not running"'
> to use ': Not running' instead.
>
> Seems a shame to lose a bunch of good catches if we can prune down the
> false-positives.  And even if this ends up missing some true positives,
> it would be a net win to catch the packages that *do* fail to check
> DEB_BUILD_PROFILES.
>
>    --dkg
>

Reply via email to