On Fri, 16 Nov 2012 23:00:49 +0200, Niko Tyni wrote: > > Think about it this way: without versioned depends, there is nothing > > to resolve the brokenness for users with squeeeze's libimager-perl > > 0.75-1 that have somehow installed libimager-qrcode-perl 0.0333-1. > FWIW, that isn't really possible as there's a major version upgrade of perl > itself between squeeze and wheezy. The squeeze version of libimager-perl > depends on perlapi-5.10.1, while the wheezy version of libimager-qrcode-perl > depends on perlapi-5.14.2, and those aren't coinstallable.
Oh, that's a good point, and I've missed it so far. Thanks! > My humble opinion is that binNMUing libimager-qrcode-perl (with > proper dep-waits) would be the minimal action too solve the RC part of > this issue, as the sid and wheezy versions of libimager-perl have the > same IMAGER_API_VERSION. No Breaks or versioned Build-Depends are needed > for working upgrades from squeeze AFAICS. Ack. > If that course of action is chosen, it would be advisable to freeze > libimager-perl in sid until the release, to ensure that any future > builds of libimager-qrcode-perl don't get accidentally compiled > with a wrong IMAGER_API_VERSION. I've added a note to d/changelog in git. > If we want protection for upgrades from wheezy/sid, the next smallest > fix would need AFAICS > - a sourceful upload of libimager-qrcode-perl 0.033-2 that Depends > and Build-Depends on libimager-perl (>= 0.90+dfsg) or something like that > - a tpu upload of libimager-perl 0.91+dfsg-3 that Breaks > libimager-qrcode-perl (<< 0.033-2) Ack. IMO that's a slight overkill. > The tpu upload could be avoided by reverting libimager-perl to > 0.91+dfsg in sid, either with an epoch or a mangled version number > (0.93+dfsg+is+0.91+dfsg or whatever.) Ouch :) > In any case, after the release, I think a proper dependency system > should be implemented like in the libdbi-perl case, and appropriate > Breaks should be added against the wheezy versions. Totally. > If there are no thinkos above (somebody please check it :), > I think it's the release team that should make the call. Right. I've now filed a binNMU bug report pointing to our discussions here, so we'll here from them, I guess. Cheers, gregor -- .''`. Homepage: http://info.comodo.priv.at/ - OpenPGP key 0xBB3A68018649AA06 : :' : Debian GNU/Linux user, admin, and developer - http://www.debian.org/ `. `' Member of VIBE!AT & SPI, fellow of the Free Software Foundation Europe `- NP: Beatles
signature.asc
Description: Digital signature