Steve Langasek wrote: > I've reviewed the diff and it looks reasonable to me, but of course > we're still blocked by the absence of t-p-u autobuilders. At present, > it seems entirely possible that the proposed d-i test candidate will > become available before the exim4 t-p-u upload will be built on all > architectures. > > If this test candidate includes an updated debootstrap which installs > libgnutls11 as part of the base system (debootstrap 0.2.42 or later), I > would suggest that exim4 could be updated directly from unstable instead > assuming there are no other changes in unstable that would be > problematic for the release. This would also let us remove libgnutls10 > from base in RC2, which is desirable in its own right. > > Joey, do you think this is doable?
We would need to coordinate the arrival of the new debootstrap-udeb in testing with the new exim -- the debootstrap-udeb would have to get in first. IIRC the original plan was to update debootstrap-udeb in one d-i release and then update exim after that release. We could try to get them in together, but there's potential for breaking rc1 for a while if we get it wrong. OTOH, rc1 will probably be broken as soon as we copy all the udebs to testing, so it should be ok to wait until that is done, and then copy in exim afterwards. -- see shy jo
signature.asc
Description: Digital signature