On 9/4/13 6:53 PM, janI wrote: > On 4 September 2013 18:13, Jürgen Schmidt <jogischm...@gmail.com> wrote: > >> On 9/4/13 5:56 PM, janI wrote: >>> On 4 September 2013 17:10, Herbert Duerr <h...@apache.org> wrote: >>> >>>> On 04.09.2013 16:13, janI wrote: >>>> >>>>> On 4 September 2013 13:59, Herbert Duerr <h...@apache.org> wrote: >>>>> >>>>> On 04.09.2013 11:15, j...@apache.org wrote: >>>>>> >>>>>> Author: jani >>>>>>> Date: Wed Sep 4 09:13:51 2013 >>>>>>> New Revision: 1519953 >>>>>>> >>>>>>> URL: http://svn.apache.org/r1519953 >>>>>>> Log: >>>>>>> update to UTF-8 >>>>>>> >>>>>>> >>>>>> Great, thanks! And kudos to Tsutomu too! >>>>>> >>>>>> But I noticed that in the l10n40 branch the *.po files of all >> languages >>>>>> are checked in amidst the main source code in the main/languages >>>>>> directory. >>>>>> Wouldn't it be better to keep all these 600 megabytes of localization >>>>>> data >>>>>> separate, e.g. in extras/ where it used to be? >>>>>> >>>>>> >>>>> In my opinion po files are just as much source as any other part of >> main, >>>>> and not something extra. >>>>> >>>>> When using the argument "extra", we should move quit a lot of modules >> away >>>>> from main, since they are not compiled into our release (at least as >> far >>>>> as >>>>> what I can see). >>>>> >>>> >>>> Yes, I agree. A lot of modules belong into ext_libraries for example >>>> avmedia, beanshell, curl, epm, graphite, hyphen, jpeg, libpng, libxml2, >>>> libxmlsec, libxslt, lucene, moz, nss, openssl, python, redland, rhino, >>>> saxon, tomcat, vigra, zlib, ... >>>> >>>> >>>> main/languages and main/l10ntools forms together with a couple of other >>>>> modules (like i18n( our integrated international part, The intention of >>>>> the >>>>> new toolset is to make a deeper integration and not push it further >> away. >>>>> >>>> >>>> In the example above with all the external modules we already have such >> a >>>> tight integration that simply moving them into ext_libraries is a >>>> non-trivial task. Our goal should not be to tighter integrate them into >> our >>>> codebase but to work towards using the off-the-shelf releases of them. >>>> Using them better with their published interfaces is a worthwhile goal. >>>> >>>> The localizations have a clearly defined tasks and they are huge. When >>>> major UI changes are underway they create heavy commit traffic. When >>>> researching the code history or when bisecting for regressions these >>>> commits cost extra time. And since we are considering to commit directly >>>> into the l10n repository from pootle this commit rate could increase by >>>> orders of magnitude. >>> >>> >>> I just wonder what would happen if we had as many developers equally >> active >>> as our translators, then the problem would be very much the same. >>> >>> I have to ask very clearly; is the opinion of the community, that .po >> files >>> is not to be considered a vital/integrated part of our source tree and >>> should be moved away ? While at the same time accepting (at least for >> now), >>> that modules as mentioned above remain in our source tree and disturb >>> developers. >> >> this question is obsolete and we all agree that the translations are >> essential as everything else in the code. We just gave you an example of >> a typical workflow and it becomes potentially better with smaller po >> files compared to 1 big sdf file per language. >> >> And the modules you have mentioned are completely different and we can >> of course discuss to remove them or store them in a different place. As >> always somebody has to do the work and it is often not only a simple >> remove. >> >>> >>> If the .po files are considered inferior, and disturbing, I will of >> course >>> not try to make an automated workflow that depend on integration. >> >> We just discuss what's better and we have at the moment different >> opinions. It's not an either or it's just seeking for the best approach >> to make all happy and to find the best solution. >> >> > I was not aware that we have different options (I as the developer do not > see these options), its really very late for that discussion. The workflow > is not up for discussion, we had that discussion based on my wiki pages > before I started programming and after several changes due to comments we > had lazy consensus.
I remember at least one chat with you where I mentioned that I would not move the po files in main directly. But because the fact that it is only an implementation detail (for me) where these files are I don't spent further time on it. > > > >> If we can define and ensure an automated workflow with continue >> integration of translation I am fine and happy with it. But if it breaks >> the build on a regular basis I am not. >> > > Changes in the .po will not be able to break a en-US build, but might break > a language build. I do however not see the difference between: if that will be the final result and workflow we will be all happy. Maybe we don't see it yet because our build process today is far away from this. > a) I as a developer introduces a build-breaker in the source (will most > likely also hit en-US build) > or > b) I as a translator introduces a translation that is a build breaker for > that language (not en-US, nor other languages) In general I agree with the minor difference that a developer can immediately fix the problem. A strange problem within help files takes probably longer or at least I don't see that it will be fixed in the same way. > > In my mindset, translator==developer, just with different tools. I sense > you see it differently. No I don't see a difference here, especially if a translator would build the lang version and did the tests immediately. But we know that this is not possible. Not today and not tomorrow with a much more simplified build env. > > And by the way I believe an automated workflow can be achieved in >> different ways. And having the po files not in main but for example in >> extras don't change this. Or maybe you have something in mind that we >> don't see or understand at the moment. But we are interested to learn >> and hear more ... >> > > Yes it can, we can f.x. just take the LO way (using the old tools, just > generating po files instead of sdf files), well I think we agreed indeed that we need something better and I don't see your point. > > I actually gave you one example why its a bad idea not to have it > integrated (different versions in extras and main or a non-existing extras). I don't get your point here > > Other workflows can for sure be implemented and will most likely work, the > only but is, you have to come up with a volunteer that wants to do it. It looks to me that you either don't want to see my point or that you are not interested in other opinions. Either we do it in your way or you don't work longer on it. Mmmh We agree in nearly every point and I of course don't understand you. We simply mentioned that we would store the po files in a separate directory, nothing more and nothing less. I hope you don't think that I am against your work or your proposal. Quite the contrary I am a strong supporter of it and I am happy that you are interested to work on this of course not easy task. I am probably not able to follow your work in all technical details but I am interested in the result and as I promised I will build your branch on Mac. Juergen > > rgds > jan I. > > >> >> Juergen >> >>> >>> rgds >>> jan I. >>> >>>> >>>> >>>> Herbert >>>> >>>> >> ------------------------------**------------------------------**--------- >>>> To unsubscribe, e-mail: dev-unsubscribe@openoffice.**apache.org< >> dev-unsubscr...@openoffice.apache.org> >>>> For additional commands, e-mail: dev-h...@openoffice.apache.org >>>> >>>> >>> >> >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org >> For additional commands, e-mail: dev-h...@openoffice.apache.org >> >> > --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
