[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Bug#744699: Frets On Fire bug report 744699



On Wed, Apr 23, 2014 at 08:58:33PM +0100, manuel.montezelo@gmail.com wrote:
> 2014-04-23 00:07 Bas Wijnen:
> >On Fri, Apr 18, 2014 at 03:02:30PM +0800, Paul Wise wrote:
> >>On Fri, Apr 18, 2014 at 2:55 PM, Gianfranco Costamagna wrote:
> >>
> >>> This is why I stopped caring about the js lintian error on sdlgfx,
> >>> put a lintian override on debian/souce symlink and I stop caring.
> >
> >This is NOT what lintian overrides are for!  Lintian tells you about
> >things that are wrong with the package.  If you stop caring, that
> >doesn't make it any less wrong.
> >
> >Overrides should only be used for cases where lintian is wrong (that is,
> >when your package is a false positive) and fixing it in lintian is not
> >reasonable.
> >
> >Adding override tags just because that makes the package lintian-clean
> >is a way of hiding problems instead of solving them.  We promise our
> >users that we don't do that.
> 
> Actually, we solved the problem for the binary package, which IMO it is what
> actually matters.

We distribute the source package as well.  For many of our programs, we
actually violate the license if we don't distribute sources.  The
license is not only about things in the binary package, but about
everything we distribute.

> lintian is right that the file does not have source, but we don't ship that file
> in the binary -- we link to the canonical file (from jquery package, or
> something like that), which has the source in clear code, and has other benefits
> (like avoiding duplication, etc).

Yes, I understand that.  But the Lintian maintainers know this, too.
And still they emit a warning about it.  If you want to argue that this
is incorrect, that's not a problem, but you should be arguing to the
Lintian maintainers (or better, to the project) that they should remove
this warning for files not in a binary package.  Overriding the warning
is equivalent to saying "I know the project thinks that I'm doing it
wrong, but they're all idiots".

As I wrote, the only time an override is appropriate is when Lintian is
mistaken and it is a false positive (and it shouldn't be fixed in
Lintian).  That is not the case here.  This is an intentional warning,
and the package rightfully triggers it.

> So, the problem is actually solved, the source-less file that lintian complains
> about is not used in any way (neither build nor run time), and thus the lintian
> error is misleading in a way, thus overriden.

No, the Lintian error means "with this source package and/or these
binary packages, we will be distributing a file without source".  That
is completely correct, and we shouldn't be doing it.  It's acceptable to
say "I have different priorities so I'm not going to fix this", but that
is not a reason to override the warning.  The bug is still there, and if
somebody is looking for fixing issues like this one, they should find
this package.  The override makes that harder.  In other words, it hides
problems.

> There are conflicting goals in the Social Contract, so waving hands and pointing
> to it every time that this topic arises on the grounds that everything that we
> offer is free, is not very helpful in my opinion.

There are some very vaque points in it, where each of us has to use
their own judgement of what to do.  But "Debian will remain 100% free"
is not one of those.  It doesn't get more exact, I would think.

And I'm not arguing that it's in the social contract and therefore has
to be the top priority.  I'm arguing that it's in the social contract,
so any violation is a real bug and not a false positive.  Therefore it
should not be overridden.

> (I am not keen on discussing this, just stating my opinion.  I probably will not
> reply to such questions in this thread).

That sounds like a good idea.  If this discussion is to be had, it
should probably be on -project or -devel.  But as long as you're not
having the discussion, please don't override Lintian warnings.

Thanks,
Bas

Attachment: signature.asc
Description: Digital signature


Reply to: