----- Original Message ----- > From: Rob Weir <[email protected]> > To: [email protected] > Cc: > Sent: Monday, September 12, 2011 12:08 PM > Subject: Re: Umbrella projects > > On Mon, Sep 12, 2011 at 11:38 AM, Joe Schaefer <[email protected]> > wrote: >> >> >> ----- Original Message ----- >>> From: Donald Whytock <[email protected]> >>> To: [email protected] >>> Cc: >>> Sent: Monday, September 12, 2011 11:07 AM >>> Subject: Re: Umbrella projects >>> >>> On Mon, Sep 12, 2011 at 10:42 AM, Rob Weir <[email protected]> > wrote: >>>> Why not just send the ballot to >>>> ooo-commits in Sumerian? >>> >>> I should think that would have to at least start out on ooo-dev-sux. >>> >>> NL development outsider here, asking for clarification...Would changes >>> to the ixn components be considered changes to the "source"? > Because >>> if it doesn't involve actual code changes I could see such a thing >>> justifying a vote on some ooo-dev-xx but then only needing lazy >>> consensus on ooo-dev. >> >> 90% of the organizational concern for releases regards their licensing, >> which I don't believe gets translated into other languages (at least > not >> without legal-discuss@ approval of the actual text.) >> >> I have no idea where ixn lives in the subversion tree, but mods that >> are committed to the tree are still mods that need to be voted on >> when it comes time to release something based on those files. >> >> My suggestion is for per-lang committers to be placed on the PPMC, >> and for those folks to conduct their votes on their per-lang list. >> Once that's accomplished, lets leave the training wheels on at first >> and ask ooo-dev@ to approve the release candidates via lazy consensus. >> Then take the whole shebang to the general@incubator list for formal >> approval by the IPMC (this step will go away when ooo graduates). >> > > Maybe someone can clear up exactly what we're talking about with a > language release. > > My understanding was we have a core code base, that has all > code-dependent i10n features in it. We also have translations, > dictionaries, etc., per language. We can build a release in English > and then require that the user download an additional "language pack" > to enable an additional language. Or we can spin off a build (more of > a new install build) to include an additional language. > > You can see this here, with the existing releases: > > http://download.openoffice.org/other.html > > So the question comes down to: what languages do we support via > officially-released install images, versus which ones are supported > via language packs? For example, today, for Uzbek, it is only > available via a language pack.
Not quite. The question is how release artifacts are based on source packages and how those source packages are managed for release. It makes no organizational difference to the ASF whether or not you use an "official release" or a "language pack", the fact is that you are using ASF mirrors for either artifact and the rules for release packages and their associated binaries apply. > > This might vary based on the timing of the translation. In other > words, we might release a new version with core languages supported, > and then enable additional languages over time as translations > complete. It would not make sense to wait for a release until all 160 > language translations are complete. > > I think it would be overkill to support this model via PPMC > delegation. OOo supports 110 languages. At 3 PPMC members per > language (for the required 3 +1's in a release vote) that comes to 330 > PPMC members. Of course, there will be some overlap, so maybe it > comes down to 200 new PPMC members or so, plus or minus 50. I'm not > sure that makes sense. > > So it is not clear that delegation to NL PPMC members really solves > the problem. We need to be having a conversation between those who > are doing the translations, those testing the translations and the > PPMC, on whether a translation is ready to release, either via > language pack or as a full install. > > Of course, if the Mentors wish to mentor 110 different NL groups on > the finer points of release management at Apache, then I don't want to > get in their way. Please. So far there is only the ja language lists in the infra queue to create: 1 !~ 110. If the ja dev community can muster three PPMC members to vote on their artifacts on their lists, that should be sufficient, once they've demonstated the general ability to oversee the legal aspects of their release artifacts appropriately. > > But I'll propose a simpler solution. We should make it easy to > nominate and approve releases of language packs and full installs > based on already approved source releases. All we need is some > indication from an expert that a given translation was ready. This > might be from a PPMC member, a Committer, or a number of Users on the > user list who have tried a pre-release language pack snapshot. We > need to rely on expertise here, expertise outside of the PPMC. But > once we decide to spin a new release, I don't think why this is not > most easily done by a vote on ooo-dev. And I'd feel much better if > the same volunteers who are building the core installs also built the > 110 language versions. It makes zero sense to have 330 different > people doing this (110 languages x 3 platforms). There is too much > scope for error. The ASF organizational mantra is about small, reversible changes whenever possible. I don't see anyone requesting the already oversized PPMC vote on 330 unknown people just because you think that's what I'm saying. The operational problem with any form of delegation at the ASF is the mustering of sustainable volunteer effort: just because it takes 3 votes from (P)PMC members to get anything done doesn't mean any 3 people who show up for per-lang service belong on the (P)PMC, must less need their own mailing list to conduct meaningful work here. As a practical matter most language efforts will end up being overseen here because there is no other mailing list with (P)PMC members on it to go. And even when there is, the likelihood of getting 3 proper +1's from (P)PMC members on just about anything is remote. All that being said, what you propose in the end isn't all that different from what I expect will happen most of the time. Binary builds of source releases should be automated/consolidated to the maximum extent possible. The only thing stopping mirrored distribution of a new binary is a prerequisite vote on the entire set of governing source code for the binary, be it a full install or a language pack.
