On Dec 11, 2011, at 9:28 AM, Rob Weir wrote: > On Sat, Dec 10, 2011 at 9:21 PM, Louis Suárez-Potts <[email protected]> wrote: >> Sorry for top posting... But yes, I was pretty much thinking along the lines >> Rob sketched out modulo TJ F. I can sound out these ideas. Political as well >> as resource issues come to mind, but I also see this as an opportunity to do >> some macro collaboration with, say, other projects engaged. In both >> localisation and ecosystem building. And as to the developer communities: >> always a challenge, but I do have many years doing just this sort of thing, >> only now there is no shroud of suspicion palling motive. >> >> (Nonsense words? iPad's spellchecker.) >> > > > Hi Louis, > > Something to consider, if you find interest in having such peer > projects at Apache. You probably don't want to do an independent, > parallel effort at IP review of the Oracle-SGA'ed contributions. That > would be painful. So it might make sense to wait until we've done > that all here, and then when we are ready to graduate, then we can go > forward with a proposal on how we deploy as one or more TLP's. Or > something like that. In general it is much easier to share > already-vetted code across Apache projects, once we've done the > initial review and cleanup work.
Well done Rob. An IP review is painful even once. Regards, Dave > > -Rob > >> -- Louis Suárez-Potts >> >> >> >> On 2011-12-10, at 20:06, TJ Frazier <[email protected]> wrote: >> >>> On 12/10/2011 19:44, Rob Weir wrote: >>>> On Sat, Dec 10, 2011 at 7:21 PM, Ross Gardler >>>> <[email protected]> wrote: >>>>> On 11 December 2011 00:13, Rob Weir<[email protected]> wrote: >>>>>> On Sat, Dec 10, 2011 at 7:11 PM, Ross Gardler >>>>>> <[email protected]> wrote: >>>>>>> On 11 December 2011 00:02, Rob Weir<[email protected]> wrote: >>>>>>>> On Sat, Dec 10, 2011 at 5:12 PM, Louis R >>>>>>>> Suárez-Potts<[email protected]> wrote: >>>>>>> >>>>>>> ... >>>>>>> >>>>>>>>> Now to the present issue. I've written that I would rather focus >>>>>>>>> here, in Apache land, on coding. But that only opens the door, as it >>>>>>>>> were, to establishing the very successful Native Language modules >>>>>>>>> either in another wing of Apache (??) or outside the Apache frame but >>>>>>>>> corresponding to it, so that QA, a key element of the NL projects, >>>>>>>>> for instance, could be tied in. Licenses, etc., would have to be >>>>>>>>> harmonised. And I'd also suggest using a simpler work medium, such as >>>>>>>>> wikis. >>>>>>>>> >>>>>>>>> I think some of this is already going on but it is not clear to me >>>>>>>>> *what* is going on or where. I'm not alone. I have received several >>>>>>>>> pings on this very question, and I'd like to move on it. >>>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> I can see several models that could work: >>>>>>> >>>>>>> All good options... >>>>>>> >>>>>>>> You hint at another option. I'm not sure it would work, but let's >>>>>>>> list it for sake of argument: >>>>>>>> >>>>>>>> 4. NL projects are individually proposed as their own podlings. Their >>>>>>>> charter would be for them to produce localizations of AOO. But they >>>>>>>> would be autonomous PMC's within Apache, with their own website, >>>>>>>> mailing lists, etc. >>>>>>> >>>>>>> Why do you feel this would this not work? >>>>>>> >>>>>> >>>>>> You have many Gaelic or Vietnamese-speaking mentors? >>>>> >>>>> Fair point, although it is reasonable to expect that many of the >>>>> people involved will be bi-lingual at least (otherwise how can they >>>>> translate). Option 4 should not be ruled out (and you didn't do so), I >>>>> was just wondering what the source of your reservations was. >>>>> >>>> >>>> So a few other ways this doesn't quite fit a podling, as currently >>>> practiced: >>>> >>>> 1) Ability to find mentors, as mentioned above. >>>> >>>> 2) Ability of our infrastructure to handle non-ASCII collaboration. >>>> We've already seen, in our small attempt to have some Japanese NL work >>>> in this project, that Roller was not allowing Japanese text and that >>>> the SpamAssassin flags every attempted post to the Japanese language >>>> list as spam. I'd expect some work would be needed in several areas. >>>> But once done, this work would benefit others who attempt something >>>> similar. So not a bad thing to try. But I'd anticipate initial >>>> challenges of this kind. >>>> >>>> 3) Technical skills needed to produce a release. To get through the >>>> ceremony of cutting a release at Apache requires someone understand >>>> things ranging from SVN tagging to GPG signing. Translators are not >>>> coders. Their expertise is on the linguistic side. They are not >>>> command-line people. You might be lucky and have someone who can also >>>> be comfortable with these things, but it would not be guaranteed. >>>> >>>> 4) The efforts can be very small in some cases. How do you get three >>>> +1's for a release if there are only 2 people in your project? >>>> >>>> 5) Growing the community of developers is hard. Once you've >>>> translated 100% of the GUI strings, then what? Translate them again, >>>> better? And then better again? Put differently, the work of >>>> translation is finite and does not give much room for growth. >>>> However, on the other, non-release side of NL projects, the outreach >>>> to users, the website, etc., there is much room for growth. >>>> >>>> 5) This creates a quasi-umbrella project. Since translations are not >>>> usable separate from the core AOO code, these other new projects would >>>> be necessarily tied to the features and the schedule of AOO, assuming >>>> they are not forking the code itself. I've heard general unease with >>>> umbrella projects at Apache. >>>> >>>> But if we are willing to dream, you could imagine a kind of umbrella >>>> project, not of code modules, but of user-facing interactions, where >>>> autonomous groups within Apache maintained localized user-facing >>>> pages, wikis, user lists, support forums, etc. TLP might be too >>>> heavy weight for this, since we have potentially many dozens of these, >>>> and their releases would consist of translated strings that are only >>>> useful when installed with AOO. The non-release activities of the >>>> project would clearly be their focus. So this is something I don't >>>> think we've seen at Apache in a TLP. (We see them in foundation >>>> projects, but this is not that). Rather than squeeze it into an >>>> existing mold, maybe it needs a new something? >>>> >>>> -Rov >>>> >>>>> Ross >>>>> >>> Idea? Some of the problems would be minimized if the Native Language >>> Confederation (NLC) as a whole became a project. Perhaps Louis could sound >>> out some folks on this? >>> >>> -- >>> /tj/ >>>
