On Sun, Jan 19, 2014 at 05:33:58PM +0100, Francesco Poli wrote: > On Sun, 19 Jan 2014 10:33:43 +0100 Antonio Terceiro wrote: > > > On Fri, Jan 17, 2014 at 11:18:46PM +0100, Francesco Poli wrote: > [...] > > > Could you please describe the chosen strategy? > [...] > > > > we will make the following change: > > > > - librubyX.Y.Z depends on rubyX.Y.Z > > - rubyX.Y.Z depends on ruby > > - ruby depends on the default ruby > > - ruby conflicts with all obsolete interpreters > > > > so this will force ruby1.8 to be removed on upgrades. > > Thanks for your reply, Antonio. > > So this will force everyone using any version of Ruby to also have the > default version, no matter what...
one bit I forgot: we also decided that we won't support more than one version in stable releases, so the 'default' version is actually the unique one. The infrastructure for having multiple versions simulaneously will stay only to support clean transitions in testing/sid. > Would the following simpler strategy work as well? > > - any supported librubyX.Y.Z conflicts with all obsolete librubyA.B.C > > This should ensure that libruby1.8 (and thus ruby1.8) gets removed on > upgrades, without forcing everyone to have the "ruby" package installed. > > Do you agree? > Or does the simpler strategy have any drawback? It does not work if you don't have packages containing C extensions (and therefore depending on librubyX.Y.Z) -- Antonio Terceiro <terce...@debian.org>
signature.asc
Description: Digital signature
