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

Re: Is Sid for broken stuff? Is it too much to ask for testing the packages?

On Thu, Dec 12, 2002 at 03:41:31AM +0100, Bernd Eckenfels scribbled:
> On Thu, Dec 12, 2002 at 03:28:41AM +0100, Marek Habersack wrote:
> > On another note - I expect the postgres maintainer to be an especially
> > cautious and responsible person, because the package he maintains is
> > depended upon by dozens of other packages and any failure to test new
> > functionality of the packaging scripts or the upstream code might cause a
> > chain reaction resulting in a lot of work for other developers
> Yes I agree, back then when I was the maintainer of Lilo I realy where very
> careful about not breaking anybodies boot sequence. This is basically the
> reason why I could not maintain the package anymore. Implementing the new
> feature requests was too much work, considering the extended care I put into
> that. And of course I still had situations which did not work well, because
> lilo is one of those packages which is very messy in terms of dependencies
> (if you consider the state of the mbr).
And it was very responsible of you to give the package away if you couldn't
cope with it because of time constraints. I'd only hope more developers
taking care of such important packages exhibited that behavior.

> For Debian we currently do not have a formal way of ensuring critical
> packages in hands which do not do too much wrong. There are only two social
> points which work quite well: 
> a) responsible base system maintainers wont give their packages away to
> people they dont trust
> b) if a package annoys too much users/developers it is getting harsh critics
> (remebered back when perl was one of those candidates?).

We might to enforce one thing, though - requirement to install the package
before uploading it. Granted, it is hard to do when the maintainer is using
plain ftp or scp, but dput (IIRC) has an option to check whether the package
being uploaded is installed - I'd vote the option to be on by default. The
same could be done for other upload clients.

> Actually I think there are not much packages which where "forced" away (at
> least I know none which was decided by RM or DPL), so this is a good sign
> and perhapos an Indication that we do not need a more formal way to punish
> DDs.
It's not about punishment, it's about developing the habit with all
developers of basic testing of their packages before uploading them. About
developing a sense of responsibility and community - because what we see now
is the lack of the latter. A person which feels tied to the community won't
ever expose other members of it to any problems which would be caused by
denial of some actions on part of that person. And, alas, this kind of
behavior seems to be plaguing Debian recently (in the past months after



Attachment: pgpNVlEsJAIkO.pgp
Description: PGP signature

Reply to: