Hi! On Mon, 2010-05-10 at 14:26:02 +0200, Raphael Hertzog wrote: > On Mon, 10 May 2010, Philipp Kern wrote: > > 1.15.7.1 seems to have a regression with relation to the handling of > > `--root'. It chroot()s into that directory nowadays and then fails to > > call the right maintainer script due to the root value still being > > prepended to the filename. > > > > This causes the following (with `--root /mnt'): > > cannot execute /mnt/var/lib/dpkg/info/nvidia-glx.prerm: No such file or > > directory > > > > strace confirms the `chroot("/mnt")' call. Am I using this parameter > > wrongly or did something break?
> The chroot call is there since long, however when the maintainer script is > called, the /mnt part (instdir) should be stripped (by preexecscript) and > for some reason it's not the case here. > > Can you submit a log generated with dpkg -D2 <your command> ? > > As a reference point, 1.15.5.6 seems to be ok according to you (info > grabbed on IRC). > > Problematic commits could be 15daa22fa94d19cc059d2755e5164db1a3a62791, > a9f8f235b90a586d99a9597fa5e7f2880ec91a98 or > ab9482eb45e27a0b0c058a2662b28b7d3642173d. They all deal with the way > "admindir" is used internally. This regression was introduced with 5050748f1a6bb0c0728f8c07f9058d545c80d7e0, it's missing assigning the path returned by preexecscript to cmd->filename. I'm testing it right now, just to make sure. We could prepare an upload with this and the other changes, Raphaƫl anything else you have pending? > I think this feature deserves a non-regression test given how seldom it's > used in daily usages by the developers. That would imply adding the infrastructure to create a chroot, which might be nice to have anyway to test dpkg w/o affecting the installed system, but still, I'm not prepared to do this right now. thanks, guillem -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org