On Sat, 10 Sep 2016, at 10:26 PM, Tanu Kaskinen wrote: > On Sat, 2016-09-10 at 21:58 +0530, Arun Raghavan wrote: > > > > On Sat, 10 Sep 2016, at 02:06 PM, Tanu Kaskinen wrote: > > > > > > On Sat, 2016-09-10 at 09:06 +0530, Arun Raghavan wrote: > > > > So unless MATE and co. are actually using it, I don't think it's a bad > > > > idea to drop it (the paprefs dep can be upgraded to latest PA with > > > > gsettings-only support). > > > > > > If the data migration is not entirely smooth for users, I want to let > > > distributions choose when to drop gconf support. > > > > Sure, making sure there's a smooth transition is important, but IMO it's > > orthogonal to supporting GConf. It's not great to have GSettings as an > > option and then push the decision of whether config should break or not > > out to distribution. > > Do you mean that it's worse to provide a gsettings option that has > glitchy data migration than to provide no gsettings option at all? I > think that depends on how high distributions set their bar. If I was a > distribution maintainer, I would stay with gconf until a smooth > migration is available. If all distribution maintainers are like me, > then I agree, we should block gsettings support until this is fully > sorted out, but if I've understood correctly, Sylvain is working on > Fedora, and his patches are less-than-perfect in this regard, which > suggests to me that Fedora perhaps doesn't care that much about the > migration problems. We shouldn't block it, but I do think we should recommend that people hold off switching until we have a clean migration path. -- Arun
