Guillem, KiBi, (It may be worthwhile taking a moment to read the bug logs of #766459 and #767999.) On Fri, Nov 07, 2014 at 8:34:49 +0100, Guillem Jover wrote: > Control: severity -1 serious > Control: retitle -1 dpkg: Correct fix breaks bogus assumptions in old debootstrap > I'd say the bug title ought to be "Correct fix makes bug in base-passwd surface." > On Fri, 2014-11-07 at 02:33:48 +0100, Cyril Brulebois wrote: [...] > > Tests performed on an amd64 host running wheezy, debootstrapping amd64. > > > > Well I bisected the archive and the last debootstrapable jessie release > > was at 20141102T221202Z; looking at the set of updated packages between > > that one and the next one, it looks like… dpkg got updated from 1.17.13 > > to 1.17.21. And unsurprisingly reverting current jessie to dpkg 1.17.13 > > makes debootstrap work again. > To stick with Santiago's earlier wording, we are now shooting the next messenger (this time dpkg, after first working hard against base-files). > What's making the breakage in debootstrap surface is dpkg 1.17.20, commit > 9ee62ecfc8937f24a82805a424564997042dd984. At least with my testing > using a patched debootstrap, and using --foreign + --second-stage to > inject a dpkg with the reverted commit. > > > A few weeks or even days before the freeze doesn't quite seem to be the > > right time to introduce (not so) subtle changes in dpkg. > > (So we need to actually freeze months in advance of the freeze… right.) > > The above commit is a *correct* fix for a very old regression, to remove > a bogus package queue dependency stage in dpkg. The breakage in > debootstrap in older versions is due to incorrect assumptions, which > where fixed correctly (not worked around, contrary to what is mentioned > on its changelog) in 1.0.56. The change was: > > - x_core_install base-files base-passwd > + x_core_install base-passwd > + x_core_install base-files > > where debootstrap was expecting that dpkg run to process base-passwd > first, even though it was passed last. This, on a call to dpkg using > --force-depends, for essential packages that do not have any kind of > dependency relationship between them, which makes any such assumption > be based on something far beyond undefined behavior. > I do agree with all the dpkg reasoning and in a way I'm grateful that dpkg made this bug surface. But really there shouldn't be any such dependency on the order of configuration of base-files and base-passwd. > > Reassigning it > > to dpkg for now; and cc-ing the release team because of things like the > > #768346 unblock request. > > I'm going to revert the commit above (only in 1.17.x, it will be kept > in 1.18.x), because it is very minimal, just reintroduces again an > unnecessary package queue stage, and such regression is acceptable if > it makes buggy bootstrappers work again. But a fixed debootstrap (and > maybe cdebootstrap if that fails too) should really be pushed to stable. > I think you might want to hold off on this revert. See https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=767999#73 and Adam's kind testing of that patch. Best, Michael
Attachment:
pgpY39nOCFPLW.pgp
Description: PGP signature