On Thu, Dec 05, 2002 at 10:02:34AM +0100, Michael Meskes scrawled: > On Thu, Dec 05, 2002 at 05:17:19PM +1100, Daniel Stone wrote: > > Secondly, ftpmasters have to add overrides for new packages, in case you > > didn't realize. That takes time: their time adding them, and our time > > waiting. We're trying to minimize the problems by ramming KDE3.1 in with > > gcc 3.2 and killing two birds with one stone. Not doing so just adds a > > I see the logic behind this, don't get me wrong. The problem just is > that the gcc movement is slowing down the KDE packages for quite some > time. After all we could have had 3.0.3 or 3.0.4 in Debian for months > now. Nope - the 3.0.x packages just were not ready to go into sid. Any package where the upgrade strategy is to purge the old and install the new just doesn't work. 3.1.x will be the first packages where upgrades from 2.2.x are clean. > > stupid amount of work to both us (the Debian KDE maintainers - Chris, > > Ben Burton, Daniel Schepler, David Pashley, Ralf Nolden, and myself), > > and the ftpmasters. > > I can understand that you want to minimize your workload, but then the > packages are there. Ralf spend quite some time on the packages and I'm > sure the others did as well. The only difference right now is that the > packages are stored elsewhere and yes, the ftpmasters don't have to add > them so far. I have not spent a huge amount of time, only a little bit of time preparing 3.0b[12] and 3.0rc[12345], as well as 3.0.3 and 3.0.3a. Karolina has been doing some unofficial packages, as has Ralf, and Ben, Daniel, David and, of course, Chris, have been spending a lot of time making sure current CVS (i.e. what's essentially 3.1), works fine; I have the packages on my laptop and they work flawlessly, thanks to the buildd a co-worker and I setup to do nothing but cvs up and debuild, 24x7. > > Until then, you'll just have to wait. More work also still has to be > > done on 2.2->3.1 upgrades before it goes into sid, but that's a > > completely moot point, because it probably won't go in for over a month, > > even if gcc 3.2 on SPARC is finally fixed. > > But then this work has to be spend anyway. Why not starting the testing > cycle now with 3.0.5 so 3.1 goes in cleanly? Not without revamping debian/control. I'm also somewhat ... uncomfortable about putting 3.0.x in. If you know why 3.1 has been held up, then you'll know that 3.0.x is quite possibly a worse choice than 2.2.x with regards to that issue. Then again, I'm just an occasional package monkey who ceased to be authoriative with my disastrous 3.0rc5 packages (the last set I made), and, to a lesser degree, long before 3.0b1 when Chris and I divvied up the packages, and I devised my exit strategy. :) d -- Daniel Stone <dstone@trinity.unimelb.edu.au> Developer, Trinity College, University of Melbourne
Attachment:
pgpqecHLnb000.pgp
Description: PGP signature