On Thu, Sep 22, 2005 at 02:52:02PM +0100, Stephen Quinney wrote: > On Thu, Sep 22, 2005 at 03:38:26PM +0200, Florian Ragwitz wrote: > > On Thu, Sep 22, 2005 at 02:11:54PM +0100, Andrew Stribblehill wrote: > > > Florian: what are you playing at? This looks like work for work's > > > sake to me. > > > > At a semi-automatic mechanism to report outdated perl modules. Basic > > approches for that were already done in pkg-perl a while ago but were > > never really used. > > > > I constantly hit problems because of that. There have been nearly 20 > > updates from, partial really outdated, perl packages. So it's kind of > > improving the quality of Perl packages in Debian, isn't it? > > It is only guaranteed to improve the quality if those updates contain > bug fixes for previous problem. Any upstream version changes which are > purely new features could easily reduce the quality instead if they > have not had time to be ironed out yet. New features can improve quality as well, if they're stable enough, of course. In that point I usually trust the upstream authors. I haven't reported bugs about new development versions (indicated by the underscore in the version number on CPAN). I thin not updating packages with new stable releases because they're not "old" enough is bad. If a module isn't stable enough and the maintainer knows about that he can tag that bug wontfix so other people are aware of that aswell. > This is not to say I don't approve of some automated service like > this. This is certainly a useful way to spot MIA maintainers for > instance. It would just be good to have some heuristics which decide > if it is worth the update yet. How should they look like in detail? > Software which is a few days "fresh" might still be quite buggy, That depends heavily on the developers release philosophy and can't be generalized, IMHO. > whereas a new version which is still not in Debian after a couple of > weeks is probably going to be as stable as we can hope for. Well, you would wait for ever if you would always wait for the next release, which is of course much better and more stable than the last one.. Regards, Flo -- BOFH excuse #335: the AA battery in the wallclock sends magnetic interference
Attachment:
signature.asc
Description: Digital signature