Thanks, I added it now. The link below should still work for the new changes, if you or anyone has additional ideas.
/Simon ons 2025-01-15 klockan 11:19 +0100 skrev наб: > Hi! > > On Wed, Jan 15, 2025 at 10:29:34AM +0100, Simon Josefsson wrote: > > My plan is to upload what's in the libidn 'wip' branch once all of > > the > > above packages are no longer in testing. Please review: > > > > https://salsa.debian.org/debian/libidn/-/compare/master...wip > > > > As you can see I'm not adding the 'Provides: libidn11-dev', as I > > don't > > think it is actually needed. No packages in testing refer to > > libidn11-dev. Things in oldstable may refer to libidn11-dev, but > > those > > packages will be upgraded to trixie versions that no longer mention > > libidn11-dev. As far as I can tell, the 'Provides: libidn11-dev' > > would > > have been necessary if we wanted to drop libidn11-dev in > > sid/testing > > before all the reverse dependencies were fixed, but I don't plan to > > do > > that since we are so close to finishing the migration. Packages > > outside > > of oldstable that refer to libidn11-dev had the bookworm release to > > migrate away from the transitional package. Does this make sense? > > I'm > > not 100% confident on this, package dependency handling during > > migrations always seems to confuse me. > I think it's still best (and most common practice, as recommended to > me) > to add a Provides: libidn11-dev ‒ there will be other packages, > both in other archives and maintained outside of distribution > ecosystems > (or, indeed, setup scripts; apt install understands Provides:), > that will keep being [Builds-]Depends: libidn11-dev (apt install > libidn11-dev), > and breaking them is unnecessarily adversarial toward the user. > The Provides: should be kept indefinitely as the end-goal of the > transition > (and the transitional package is just the only way to get to this > state). > > So, /Suggests:/iProvides: libidn11-dev > is IMO the best course of action, regardless of the state of the > archive > (but, of course, the rdeps should still switch to libidn-dev). > Beside that, looks fine to me. > > Best,
signature.asc
Description: This is a digitally signed message part
