Christopher C Chimelis writes: > > Did we ever decide on what to do with glibc as far as the lchown() issue? > The discussion was a bit lengthy so I missed some of it, but never got a > resolution. Was it decided to continue on the way it is or change it > somehow? I propose the patch below, it apply on the unpacked source glibc package, there are corresponding binary deb packages on http://lhpca.univ-lyon1.fr/~lprylli/glibc/. I strongly suggest we take this one or an equivalent one that guarantee at least the same good properties as far as the (l)chown issue is concerned (hope I am correct with these assertions): - follow the same semantics that most(all?) other ports - should work perfectly with 2.0 broken kernel,2.0 fixed kernels, >= 2.1.9? kernels (people running unstable kernels between 2.1.0 and 2.1.9? will still have the problem unless you had one test in the patch). - should work perfectly with any old or recent, patch or unpatched version of dpkg, tar, chown, ... - should work with any combination of the the above entities. - upgrading to this library is safe from any start point. At the same time, I suggest we put a predepends in dpkg, tar, and maybe fileutils, on a recent enough library (that includes lchown) to guarantee nice upgrades. Of course maybe some other people should check that my patch is correct and the deb work everywhere. Regards, Loic
Attachment:
glibc.patch
Description: Binary data