On Sun, Jun 28, 2009 at 8:35 PM, Jonas Smedegaard<d...@jones.dk> wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: RIPEMD160 > > On Sat, Jun 27, 2009 at 09:23:05AM -0400, Jan Muszynski wrote: >>The version of cups in sid (1.3.10-5) has a recommends on >>ghostcript-cups. ghostcript-cups has a dependency on cups, but it's >>unversioned. Does the cups dependency need to be >= 1.3.10-5 ? For >>anyone finding this note that cups is currently stuck in sid >>apparently waiting on the krb5 transition (which seems to be waiting >>on some of the architectures other than i386/amd64). > > Above is a bug in cups (or the helper packages more directly depending > on raster filters. The maintainer of cups is fully aware of the need > for the new ghostscript-cups package, as that maintainer reqested that > very change.
Right, and the new version of cups does, indeed, have a recommends on ghostscript-cups. So that part of the equation is taken care of in the sid version of cups. The question I have (which cannot be fixed in cups itself) is should the dependency on cups within ghostcript-cups have a version attached to it? ghostscript-cups needs cups in order to function. Fine, that's why it has the dependency. But will it function with any version? Or does it need the version that's currently in sid? If it needs the version currently in sid then the dependency on cups should be versioned. That's all. This isn't a huge deal since once cups itself finally moves down to testing it will self-correct. But in the meantime it will be problematic. Someone upgrading ghostscript will have no way of knowing that the new ghostscript-cups package exists (which was why I suggested a suggests dependency). Someone that finds the new package won't be aware that they need the version of cups in sid rather then the version they already have installed (assuming they run testing. If they run pure sid they're already in synch). Since ghostscript-cups obviously depends on cups (as the dependency correctly states) I just feel that it should version the already existing dependency to indicate it won't work with the version of cups it finds in testing. The version found in sid appears to work just fine - so no bug in that version of cups. > > >>Ghostscript itself should possibly have a recommends, or at least a >>suggests, on ghostcript-cups. > > No: Ghostscript benefits in no way from the contents of > ghostscript-cups. But the contents of said package were, prior to this point, part of ghostcript itself. So in a twisted sort of way you're stating that ghostscript doesn't benefit from itself :) > > I will now close this as it is not a bug in ghostscript. OK - can you reassign to ghostcript-cups? :) > > > Kind regards, and thanks for your report, anyway :-) You're welcome. At least if anyone runs across the problem and finds this bug listing they should be able to figure out what to do. (Which is the biggest reason I'm doing this). Hopefully the cups version in sid will make it down to testing sooner rather than later, which will make this discussion pretty much moot. -- To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
