On Tue, Aug 10, 2004 at 10:23:16PM +0200, Jose Carlos Garcia Sogo wrote: > On Mon, Aug 09, 2004 at 06:55:52PM -0700, Steve Langasek wrote: > > On Mon, Aug 09, 2004 at 10:32:52PM +0200, Adrian Bunk wrote: > > > Could anyone explain how this could work in practice? > > > > > Since in both cases the version would be higher than the one currently > > > in unstable, a package compiled with 1.0 would be allowed to use 0.7 . > > > > > You could add versioned conflicts for current users of libgpg-error0 in > > > unstable and update the build dependencies of such packages using > > > libgpg-error0. > > > > This works because so far, none of the packages rebuild against > > libgpg-error0 *use* libgpg-error0 -- it is a spurious dependency. > > > > Jose is right that the packages which currently are blocked by the new > > libgpg-error0 in unstable should be uploaded with a fixed libtool; but > > he is wrong if he thinks this excuses introducing such major changes > > while we are preparing for a release. > No, this is not my excuse for the upload. I simply decided to do the > upload two days before realizing that it was base, and when I did the > real upload, changing the priority, I only think that I had my mental > ok for uploading the package (without thinking on the implications for it > being base). > I'm very sorry for all the inconvenience I have caused. These things happen sometimes, and it's clear that this was nothing deliberate on your part. As always, the important thing now is to fix the bugs, so thank you for your help with that. > > Jose, please re-upload > > libgpg-error 0.7 with either an epoch or a "1.0-really-0.7" version > > as Goswin has suggested. Or, if you don't have time for this, please > > indicate which of these you prefer and I'll NMU according to your > > wishes. > Ok, I'll upload it tonight, so it's ready for tomorrow katie run. Ok. Please note that my patch is incomplete, as it generates ">= 1:0.7" for the shlibdeps. There has also been discussion among the release team about whether this change could be allowed into testing after all, because the changeset is so small, but a decision hasn't been made yet. I believe reverting the change is best for now, even though we may still need to change strategies for sarge. Thanks, -- Steve Langasek postmodern programmer
Attachment:
signature.asc
Description: Digital signature