Re: Work-needing packages report for May 25, 2001
>> Sam Couter <sam@topic.com.au> writes:
> How long do orphaned packages survive? I ask because no one seems to want
> debauch, it's been orphaned for a long time, and there are better
> alternatives available.
My opinion is that we should not have orphaned packages. At all. The
Quality Assurance group should be doing that: providing Quality
Assurance for the release, instead of maintaining old stuff which noone
in the project is interested in. Since that's unreasonable and
unpractical, and since not all packages are created equal, a line
should be drawn (ok, two lines): important packages (that is, standard
and up) should be maintained by the QA team (this does fall into QA's
job description if you take into account the definition of "standard");
other packages should stay unmaintained for a reasonable ammount of
time, after which they get moved to the withdrawn pool[0]. Maintaining a
package also means *using* it and *noticing* when it breaks, not
waiting for an RC bug report to come in. In a manner of speaking, we
/do not/ want bug reports. When the bug report kicks in, it's already
too late. This is also not practical: you can't ask a maintainer to be
fully aware of when and how a package breaks (think X), but you want
him to have the required expertise to fix it (and part of that
expertise comes from actively maintaining a package). If a package is
so great that it won't break (think document) and does't need active
maintainance... why hasn't someone picked it up already? Just
yesterday I got two mails regarding packages I don't maintain anymore
because I'm listed as the maintainer for them on some potato CD. Since
I *did* work with the packages when they were still useful to me, it
took me 15 minutes to answer both mails.
Anyway, the last time this discussion took place we didn't reach any
agreement other than preserving the status quo.
> Should bugs be filed against ftp@debian.org to remove packages that
> have been orphaned longer than some number of days? I believe this
> has been done in the past as a once-off type of thing, but should it
> become an automatic thing with each mailout of the wnpp report?
As other have (or will) pointed out, the problem with this is that
removing a package is easy (and now easier than ever). Getting it back
in is not that easy. I do admit automatic filing of this kind of bugs
is a Very Bad Thing(TM). I wouldn't have a problem with spam^Wsending
the -qa list a mail saying "hey guys, these packages are *really* old
and don't have a maintainer, is there a **REAL** reason to keep them in
the archive?" But I'm afraid I know the reply each and every one of
those mails would get... (hi Adrian ;-)
[0] I don't like the idea of getting completly rid of a package,
specially if there are changes between the version in the last
release and the withdrawn version.
--
Marcelo | "Go ahead, bake my quiche"
mmagallo@debian.org | -- Magrat instructs the castle cook
| (Terry Pratchett, Lords and Ladies)
Reply to: