On Tue, Jul 26, 2011 at 11:20:10AM -0400, Tom Lane wrote: > Toshio Kuratomi <a.bad...@gmail.com> writes: > > On Tue, Jul 26, 2011 at 10:37:43AM -0400, Tom Lane wrote: > >> Michal Hlavinka's solution of explicitly testing for the old sysv init > >> script seems like a win from here, since I don't intend to continue > >> packaging that. Anyone have an objection to that approach? > > > Yes, I object. As Michal said in his post, the %post that he uses is > > problematic if someone has installed a package with sysv init scripts for > > that service. > > That argument seems like a straw man, considering that the file to be > tested for was provided by the previous version of mysql-server, and > that the test would only be made when we know we are upgrading (not > freshly installing) mysql-server. It's hardly likely that anyone was > providing a conflicting version of it. Moreover, what's the downside > if someone did? His sysv-based boot configuration would get migrated > to systemd. Not exactly fatal, I think. > Alternate init systems are currently allowed within Fedora. The init scripts may come from subpackages of the service-providing package or they may come from entirely separate packages so that they can bootstrap that init system and so that they can be done without causing the service-providing package maintainer to do work for things that are not the default init system.
With that in mind, your assumptions are wrong. > > Please read the link I posted to Ville's message instead[1]_. > > Thanks, but I do not intend to make it my job to do extensive testing > of someone else's scriptlets. This work should have been done by the > systemd team before foisting a poorly-thought-out upgrade process on > the rest of us. > Well, then you'lll just have to use the scriptlet that is in the packaging guidelines as it's the only one that we've tested. > On balance I still think that Michal's solution is the least risky. > None of these solutions are perfect, but that one is the least likely > to fail over time. > Let me say it again, a little differenttly then: Michal's solution has known breakage. Ville's proposal has no known breakage (if you follow the correct recipe that he gives) at this time. Ville's method is an extension of the method present in Michal's solution with additional logic to not break in the situations that Michal's is known to break under. If you still consider Michal's solution the least risky knowing that it's broken and that what's essentially a bugfix to it exists, then I don't know what to tell you. -Toshio
pgpi1VT8A5b0w.pgp
Description: PGP signature
-- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel