* José Luis Tallón <[EMAIL PROTECTED]> [2008-04-30 22:25:10 ART]: > Marc 'HE' Brockschmidt wrote: > > Restricting a package to some architectures because you are neither > > able nor willing to fix the problem on other archs is not a fix for this > > bug. It's a workaround, nothing more. > > Indeed it is a workaround.
But one that bites our users, as have several people tried to point out to you. > However, I think having 0.5.2-1 in the archive (even though restricted > to two arches) is better than having 0.5.1 which does FTBFS. Others disagree here, and what's more important, release managers do disagree, especially since there is no FTBFS for 0.5.2-1. At least not on powerpc, see the build log I managed: <http://rhonda.deb.at/debian/tmp/lcdproc_0.5.2-1.1_powerpc.build.bz2> > The alternative would be (although we are probably too close to release > to try it) to try and build it for all arches and await confirmation > from users that it does work (or does not). At the time of your writing there was more than enough time for that... It's different now, it might get tricky to get the update of 0.5.2 into testing. Especially since you managed somehow to have a quite big gap from 0.4.5 to 0.5.2 now sitting around between testing and unstable. > It happens that I only have confirmation of lcdproc working properly on > these right now (for the current version). It's very rare that you receive messages about stuff to be working on any architecture in specific. Especially when you forbid them to test it conveniently by forbidding the package to build on those architectures. Unfortunately I can only report that 0.5.2-1 builds fine on powerpc (once it's added to the Architecture: line) but I can't report that it works when running because I don't have the proper hardware to test it. But you are limiting people who would actually be able to test it wether it works with the required hardware absolutely to those who are able to rebuild the package themself - which is a quite unpleasent thing. > > Also, setting the architecture header is not enough, someone would also > > need to remove the old binaries from unstable. > > Yes. I just agreed to this a couple hours ago at #400066. ... and you never did. > If you would sponsor it, we can certainly upload an "architecture: any" > version of lcdproc --- please remember that I have no access to any > porting machine and am thus unable to test compilation on other arches. That's what the buildd network is there for. > It just took me some three weeks to have some packages uploaded, since > Amaya was too busy, and my other sponsors are basically MIA. Did you ask on the -mentors mailing list? On the -mentors IRC channel? It shouldn't be hard to really find someone willing to upload a package for a RC bug, and that includes myself. I'm willing to upload the package widening its Arch list to any. > I can prepare the updated package and have it ready in some hours, if > you agree. Did you do that at some point in the past? There is no status for that in the bugreport, so I'm unsure. So long, Rhonda -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
