On Fri, Jun 09, 2017 at 09:32:55AM +0200, Lucas Nussbaum wrote: > On 08/06/17 at 22:25 -0300, Antonio Terceiro wrote: > > Can anyone think of a good reason to not go straight to 2.5, starting > > ASAP after the release with a pre-release ruby2.5? > > Before the official release, there's probably no guarantees about ABI > stability. Couldn't we end up in a difficult situation if we package a > pre-release of 2.5 with libruby-2.5.so.2.5, and the ABI later changes in > an incompatible way? > > (Maybe that's a risk that we should take anyway, and deal with it with > binNMUs if it happens) I think we can start testing rebuilds with ruby2.5 in experimental way before we start the actual transition in unstable. We can leave starting in unstable to some time just after upstream enters in feature freeze. On Fri, Jun 09, 2017 at 10:30:49AM +0200, Christian Hofstaedtler wrote: > * Antonio Terceiro <terceiro@debian.org> [170609 03:25]: > > We are now more or less 6 months from the release of Ruby 2.5, and I was > > thinking whether it makes sense for us to transition to Ruby 2.4 and > > then transition again to Ruby 2.5. > > [..] > > > Can anyone think of a good reason to not go straight to 2.5, > > I think avoiding 2.4 because it would only be extra work for us with > no benefit does make lots of sense. > > > starting ASAP after the release with a pre-release ruby2.5? > > On this I share Lucas' concerns about ABI stability, plus Ubuntu may > end up shipping a pre-release 2.5 then. Why not wait until 2.5 is > out before sending it to unstable? Yes, we can do that. To me the most important change is not doing a 2.4 transition. > I can see how it'd be nice to test with pre-release versions, but > maybe this is better done elsewhere (similar to what is done with > Haskell AFAIK?). Yes.
Attachment:
signature.asc
Description: PGP signature