> > My understanding is that upstream is very unhappy with the fact that > Debian is shipping an old version, as it generates support requests for > something that they don't want to support (3.0.5 was released on > 2008-07-11). >
Right. One of the points we have drive home to the users is that if they have a problem with the current version they have to try the latest available version before reporting that bug to either BTS or upstream. This again intersects a bit with the regression tests I mentioned before. Normally, I'd expect that users DO NOT have the infrastructure to install the latest version. But if they pop up their problem on debian-science then one of us can test it and tell them if the problem goes away by compiling/installing the latest release. If the problem persists in the latest release, then it gets reported upstream. This happens to me all the time with gfortran. I'll come across a code that goes boink when I use the current version of Debian. However, the gfortran devs only care if the problem is reproducible on the svn head. Due to lack of man power, they don't backport their patches. So, I ask on their mailing list if the problem is fixed in subsequent version. If the answer is yes, then I am happy to upgrade my gfortran. If the answer is no, then I report the bug. Users do not want to install the latest version without knowing if their problem is fixed in the latest version. But if they know that the problem is fixed, then they make every effort to get that version. > Removing the package from unstable doesn't prevent you from working on > the package. It's just a way to clean up Debian. It will be very easy to > re-upload when you will have something that builds in i386 and amd64 > (though it might be better to upload to experimental, as I doubt that > you will have something in a releasable state before a few months). I agree. Actually removing the package might do some good in this case. We can concentrate on just i386 and amd64 and worry about other architectures later on. This might actually speed up things a bit. > > As for the strategy of working on the 3.X release or on the 4.X release, > I don't think that we should try to release something which is not > closer to the latest upstream release than 3.0.5. Yes. The software released in Debian should be close to the latest 4.X release. I am thinking of having the intermediate versions somewhere in a private repository. -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
