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

Re:%20Re: Debian Maintainers GR Proposal

>> So here was my practical conclusion: I did send a bug report, useless
>> during months, and that bug report was used to argue that the package
>> is
>> broken and unkaintained and to remove it. Conclusion: reporting on a
>> un-maintained package is something dangerous.
>Hm, what was the severity of the bugs you are thinking about?  Where the
>packages removed due to a couple of unfixed important bugs?

There was two kinds of bugs in the package:
- the one I reported, with a patch (segfault on some cases introduced by
  modern TeX uses)
- bugs about Debian (version of the standards, version of dh_*, etc)

Of course, the bugs about Debian were regarded as very serious, and were
the reason for being a candidate to removal. From a user point of view,
it sounds like crazy.

> I think that Debian users are generally better served when the number of
> available packages is limited, but their quality is high, instead of a
> large number with often questionable quality.  This is particularly true
> for software which is hardly maintained upstream; I guess dvidvi is an
> example for that:  Since I started using TeX seriously, around 98 or 99,
> I've never been given advice to use it

Neither do I. But for some precise jobs, dvidvi is the only tool to be
available, even if those jobs are not frequent ones. The ps* tools are
not aware of the real page number in a TeX document, while dvidvi is.
When you want to isolate "the third page 1 of the document" there is
nothing out of dvi* tools.

> (I use ps* or pdfpages instead).
> If it's being maintained badly within Debian, is there much use in
> keeping it?  Those with old installs can keep their old version, those
> with new ones should probably rather learn those tools that are
> currently developped.  The only problem is if a long-term user gets a
> new machine...

I strongly disagree. There are two ways required to improve quality:
- having a strong subset, well maintained, quickly evolving, packages
  (in that category you'll find the kernel, or openoffice, or firefox),
  those packages are associated to software too unstable which need to
  be updated really often, and which need to be coherent and fully
  tested together
- having a stable subset, those tools which evolve slowly, often because
  there is no need to update, but have been around for a long time

If you remove the second category, then (for example) a Makefile writen
a few years ago may not work any more. This is not a good indication if
you perpetually need to rewrite everything.

Of course, when there is a conflict (e.g. we cannot maintain such very
old package without breaking some brand new one) the conflict need to be
sorted out. But in a lot of cases, packages tend to be dropped because
there in no more maintainer, while they create no real problem.

Another case come back in my mind: pandora. Those fonts have been
available with TeX since years and years. They have been removed from
Debian/main for good reasons (wrong license: free for non commercial use).
In my mind, in such a case, it should be mandatory to move the
corresponding software in Debian/non-free, instead of dropping it.
It is often not done because the maintainer do not have time enough to
do that, it can be sorted out by a DM who would handle that.



Reply to: