Le Mon, Oct 05, 1998 at 01:57:46AM -0400, Adam P. Harris écrivait: > C'mon, let's be realistic. Introducing a new perl right now would > basically blow a 1998 stable slink out of the picture (as would PAM, > of course), IMHO. It's not just a question of re-uploading 35 > packages, but also of testing, and tracking down interactions. Maybe > I'm too pessimistic, but it seems like a big problem to me.
Yes, You're too pessimistic :), what problems of testing and interactions do you mean ? As far as I know, recompiled modules work out of the box. Correct me if I'm wrong but I dont think. > It would be much preferable to wait until post-freeze, and then, > w.r.t. the libraries, either have perl conflict with the older > versions of things, or else provide a big perl5.05 "profile package" > or some such external mechanism to ensure the smoothness of the > transition. Yes, if we do nothing for slink then we will have to make perl5.005 (and all perl package after 5.005) conflict with a list of 35 package with precise version. It seems to me ungraceful... the other solution would be to add a conflict line for actual lib*-perl packages but then they need to be modified and uploaded, so why not just upgrading ? > If it's a big issue, why can't we plan on having a Debian 2.2 release > very quickly, i.e., freeze again in January or Feb? Our main goal is Why should it be a big issue ? It's just a matter of saying "ok from now we will use perl5.005" so lib*-perl package will depend on it and we have to recompile them. It's not like PAM where interactions with other packages are important. > an integrated, stable system. We shouldn't give into fetish for the > newest versions, at the expense of stability. Perl 5.005 is stable. Cheers, -- Hertzog Raphaël ¤ 0C4CABF1 ¤ http://www.mygale.org/~hra/
