Re: Bug#167886: marked as done (tux-aqfh-data: should replace older versions of tux-aqfh )
On Mon, Dec 02, 2002 at 12:36:34AM +0100, Matej Vela wrote:
> Colin Watson <cjwatson@debian.org> writes:
> > On Sun, Dec 01, 2002 at 05:33:18AM -0600, Debian Bug Tracking System wrote:
> > > tux-aqfh (1.0.14-2) unstable; urgency=low
> > > .
> > > * QA upload.
> > > * Make tux-aqfh-data replace tux-aqfh (<< 1.0.14-1) to avoid conflicts
> > > on upgrade. Closes: #167886.
> > > * Remove undocumented(7) symlink for tux_aqfh(6). Its lack has been
> > > reported as #171283; add Lintian override.
> > > * Conforms to Standards version 3.5.8.
> >
> > Why the Lintian override?
>
> I was under the impression packages with Lintian errors are a big no-no.
In this case it's partly a Lintian bug. The link-to-undocumented-manpage
warning should be removed altogether, as policy has been amended to no
longer recommend the use of undocumented(7).
As for leaving the binary-without-manpage error there, please do! It's
not completely out of order to have packages with Lintian errors, in the
same way that it's not out of order to have packages with bugs; we just
have to acknowledge that they *are* bugs. Overriding the error isn't
such an acknowledgement, it's saying that Lintian is wrong in this
particular case and should be quiet.
Basically, the package still has the bug that the man page isn't
present, so Lintian's output should continue to mention it. I've written
man pages for packages more than once after noticing a complaint from
Lintian.
> Overriding link-to-undocumented-manpage was acceptable:
>
> W: tux-aqfh: link-to-undocumented-manpage usr/share/man/man6/tux_aqfh.6.gz
> N:
> N: Symbolic links to the undocumented(7) manual page may be provided only
> N: when a bug has been filed that no manual page is available. If you
> N: like, you may report the bug yourself, and add an override for this
> N: warning in your package.
> N:
Bleh, I disagree with that text. :-) Shaleh was generally consistent in
saying that overrides should only be used for cases where an exception
needs to be made to an otherwise correct general rule, and not to hide
Lintian bugs or package bugs. I don't know what Joy thinks, but this has
certainly been the guideline in the past.
(I wonder if that text was my fault ... if so, oops.)
> If we override binary-without-manpage for reported bugs, it'll be easy
> to spot unreported ones.
But we won't see the reported ones in the overall list that way.
> > If there's no man page, the error should stay there without being
> > overridden so that it appears on summaries like
> > http://qa.debian.org/man-pages.html.
>
> How did it work in the undocumented(7) days then?
People generally didn't override the warning.
Cheers,
--
Colin Watson [cjwatson@flatline.org.uk]
Reply to: