On Sunday 04 March 2007 02:05, Ciaran McCreesh wrote: > On Sun, 4 Mar 2007 01:51:39 +0900 Jason Stubbs <[EMAIL PROTECTED]> > > wrote: > > There were two separate specifications - glep42 and multiple > > repositories - that should have been discussed seperately. On a > > seperate thread, Marius said something to the effect of "specs are > > much easier to extend than to alter". Having read that, I think we > > were both wrong - specification of a repository should probably have > > been left out completely until repositories had been hashed out. > > Which is what I was pushing for all along. I was trying to leave > multiple repositories entirely out of it -- despite Paludis making use > of them -- because the GLEP had nothing to do with multiple > repositories.
I don't remember the specifics, but I remember that there was something that didn't seem to go along with our vision. But yes, I do remember you pushing for keeping multiple repositories out of it. In general I try to look at everything as my fault (ie. what could I have done different?) and in that case I probably should have moved to remove whatever it was that didn't sit right rather than pushing to have it adjusted to my vision. > > To sidetrack just a little more, I think this illustrates one of the > > reasons why having PMS (aka EAPI-0 spec) completed is so important. > > Not only would it allow for Gentoo package manager(s) to be > > interchanged/replaced, it would provide a incontrovertible context > > for discussion of new features. > > There aren't going to be any new features in PMS. The only package > manager changes that we want to come about as a result of it are bug > fixes. Yep and that's a good thing. > Incidentally... Side note on PMS vs EAPI-0: the way PMS is written, > it's deliberately very easy to integrate EAPI-1, EAPI-2 or whatever > into the document. Consider PMS to be a document that is capable of > holding all EAPIs, with EAPI-0 being the only one that's actually there > for now. Once EAPI-1 is agreed upon, it can be added to PMS rather than > having to be a whole new document. That also sounds like a good thing as it gives new ebuild authors a single authoritative source on what to expect from a package manager. Although EAPI-0 will still be defined, even if it is only as "revision XYZ of PMS". Also, as a leading dev to a (for a? on a? i've spent too long in Japan :/) "not an official Gentoo project" package manager, I hope you realize the danger of not having explicit versions of the document. Take, for example, the lack of acceptance of some changes to the dev guide that have been somewhat controversial... > > This is really irrelevant. It's not matter of "if" he gets access but > > only as to "when". After the initial work is done and the team is > > ready to go public all his "noise" will come out. I can only think of > > two choices here: 1) whether you and he both continue to be visceral > > or instead try to build a good working relationship; and > > 2) whether you discuess any issues with the spec now or when it goes > > public. > > That's a fairly big difference. If it's later on, there won't be lots > of holes that we know are there that he can use as some kind of twisted > proof that PMS sucks. That's a "yep" again to it being a fairly big difference although I won't back your justifications. It's something you as a team and ultimately Stephen needs to decide. Either way, I'm just reminding all that you're not preventing Brian from having a say. Anyway.. Unless your reply has either something that I don't agree with or that is really exciting, I'll let you have the last say. (Why is it that those that are technically minded always want to have the last say? ;) -- Jason Stubbs -- gentoo-dev@gentoo.org mailing list
