On Sun, 2022-09-18 at 13:20 +0200, Ansgar wrote: > On Sun, 2022-09-18 at 12:11 +0100, Luca Boccassi wrote: > > On Sun, 2022-09-18 at 11:42 +0200, Sven Joachim wrote: > > > This does not quite work as intended, because debootstrap's > > > simplistic resolver only ever looks at the first alternative in > > > dependencies, and so usrmerge gets installed anyway. See > > > #768062, > > > for instance. > > > > That doesn't match what I have seen while testing it, and what I > > see > > even now in a sid chroot where usr-is-merged is pre-installed: > > Did you try debootstrap with init-system-helpers 1.65.2 in unstable? > I could reproduce the problem; "debootstrap --print-debs unstable > unstable https://deb.debian.org/debian"; or similar should be > sufficient > to show the problem. > > I wrote a possible patch for debootstrap in [1], but being > debootstrap > we might need to have it in stable as well. Maybe someone has other > ideas as well. > > Ansgar > > [1]: > https://salsa.debian.org/ansgar/debootstrap/-/commits/exclude-usrmerge Indeed, that test was in an already installed chroot. This is very strange as I am sure I had tested with a local reprepro with the new packages when testing the debootstrap changes, and I recall that it was working as intended. Evidently I was mistaken. Could you please fire a MR with those commits to the deboostrap repo? The most important part is setting up the chroot correctly, the additional dependencies are an annoyance but can be removed post-fact and shouldn't break things, so maybe it's fine to update it only in unstable for now? -- Kind regards, Luca Boccassi
Attachment:
signature.asc
Description: This is a digitally signed message part