Fwd: New module proposal: Clutter core
This was by mistake sent to only gnome-i18n - forwarding now also to devel. (Thanks, Christian Rose) -- Forwarded message -- From: Ask Hjorth Larsen Date: Mon, Oct 11, 2010 at 11:16 AM Subject: Re: New module proposal: Clutter core To: Kenneth Nielsen Cc: GNOME i18n list On Mon, Oct 11, 2010 at 10:54 AM, Kenneth Nielsen wrote: > 2010/10/10 Petr Kovar : >> Hi! >> >> Matthias Clasen , Thu, 7 Oct 2010 07:55:42 -0400: >> >>> On Thu, Oct 7, 2010 at 7:51 AM, Vincent Untz wrote: >>> >>> > I think the right solution is to improve l10n.gnome.org so that it can >>> > interacts with other platforms (like transifex) where there is a team >>> > control. >>> > >>> > Alternatively we can also investigate something similar to where >>> > transifex is heading to: host the po files on l10n.gnome.org and add >>> > some autofoo magic that would download the po files from there during >>> > 'make dist'. >>> >>> I think moving toward transifex or a similar distributed approach for >>> translations is the way to go. >> >> When it comes to consistency in translation (which matters), not >> distributed approach, but centralized effort of community is the key to >> success. That's what I learned from several years of translating FLOSS >> projects. And I guess that other translators observed that as well. > > I agree. From the point of a translation coordinator. The best choise > is to have all translations in one place that he/she controls. > Alternatively, when that fails the second best options is to have all > the translations centered in as few solutions as possible. And there > are important things to look for when determining which to use, but > I'll dive into that in the separate thread that has been stated. > > \Kenneth And from the point of view of a translator grunt who just wants to convert as many strings as possible with as little fuss as possible, it's much easier if you can just get those sodding files from somewhere central :). Experienced translators will find it annoying to have to go somewhere else, and to new translators it will even be a barrier. The ability to (so far) get all the GNOME strings from one place should not be underestimated. Regards Ask ___ gnome-i18n mailing list gnome-i18n@gnome.org http://mail.gnome.org/mailman/listinfo/gnome-i18n
Re: New module proposal: Clutter core
2010/10/11 Christian Rose : > 2010/10/11 Ask Hjorth Larsen : >> On Mon, Oct 11, 2010 at 10:54 AM, Kenneth Nielsen >> wrote: > [...] >>> I agree. From the point of a translation coordinator. The best choise >>> is to have all translations in one place that he/she controls. >>> Alternatively, when that fails the second best options is to have all >>> the translations centered in as few solutions as possible. And there >>> are important things to look for when determining which to use, but >>> I'll dive into that in the separate thread that has been stated. >>> >>> \Kenneth >> >> And from the point of view of a translator grunt who just wants to >> convert as many strings as possible with as little fuss as possible, >> it's much easier if you can just get those sodding files from >> somewhere central :). Experienced translators will find it annoying >> to have to go somewhere else, and to new translators it will even be a >> barrier. The ability to (so far) get all the GNOME strings from one >> place should not be underestimated. > > FWIW, if you and Kenneth are only replying to the gnome-i18n list, you > are only preaching to the choir. We, translators, already know this. > > It is on desktop-devel-list, where new module proposals are being > proposed and discussed by all contributors, where such important > points should be raised. Ah yes, we will re-raise these concerns in the separate thread and on the appropriate list. \Kenneth > > > Christian > ___ gnome-i18n mailing list gnome-i18n@gnome.org http://mail.gnome.org/mailman/listinfo/gnome-i18n
Re: New module proposal: Clutter core
2010/10/11 Ask Hjorth Larsen : > On Mon, Oct 11, 2010 at 10:54 AM, Kenneth Nielsen > wrote: [...] >> I agree. From the point of a translation coordinator. The best choise >> is to have all translations in one place that he/she controls. >> Alternatively, when that fails the second best options is to have all >> the translations centered in as few solutions as possible. And there >> are important things to look for when determining which to use, but >> I'll dive into that in the separate thread that has been stated. >> >> \Kenneth > > And from the point of view of a translator grunt who just wants to > convert as many strings as possible with as little fuss as possible, > it's much easier if you can just get those sodding files from > somewhere central :). Experienced translators will find it annoying > to have to go somewhere else, and to new translators it will even be a > barrier. The ability to (so far) get all the GNOME strings from one > place should not be underestimated. FWIW, if you and Kenneth are only replying to the gnome-i18n list, you are only preaching to the choir. We, translators, already know this. It is on desktop-devel-list, where new module proposals are being proposed and discussed by all contributors, where such important points should be raised. Christian ___ gnome-i18n mailing list gnome-i18n@gnome.org http://mail.gnome.org/mailman/listinfo/gnome-i18n
Re: New module proposal: Clutter core
On Mon, Oct 11, 2010 at 10:54 AM, Kenneth Nielsen wrote: > 2010/10/10 Petr Kovar : >> Hi! >> >> Matthias Clasen , Thu, 7 Oct 2010 07:55:42 -0400: >> >>> On Thu, Oct 7, 2010 at 7:51 AM, Vincent Untz wrote: >>> >>> > I think the right solution is to improve l10n.gnome.org so that it can >>> > interacts with other platforms (like transifex) where there is a team >>> > control. >>> > >>> > Alternatively we can also investigate something similar to where >>> > transifex is heading to: host the po files on l10n.gnome.org and add >>> > some autofoo magic that would download the po files from there during >>> > 'make dist'. >>> >>> I think moving toward transifex or a similar distributed approach for >>> translations is the way to go. >> >> When it comes to consistency in translation (which matters), not >> distributed approach, but centralized effort of community is the key to >> success. That's what I learned from several years of translating FLOSS >> projects. And I guess that other translators observed that as well. > > I agree. From the point of a translation coordinator. The best choise > is to have all translations in one place that he/she controls. > Alternatively, when that fails the second best options is to have all > the translations centered in as few solutions as possible. And there > are important things to look for when determining which to use, but > I'll dive into that in the separate thread that has been stated. > > \Kenneth And from the point of view of a translator grunt who just wants to convert as many strings as possible with as little fuss as possible, it's much easier if you can just get those sodding files from somewhere central :). Experienced translators will find it annoying to have to go somewhere else, and to new translators it will even be a barrier. The ability to (so far) get all the GNOME strings from one place should not be underestimated. Regards Ask ___ gnome-i18n mailing list gnome-i18n@gnome.org http://mail.gnome.org/mailman/listinfo/gnome-i18n
Re: New module proposal: Clutter core
2010/10/10 Petr Kovar : > Hi! > > Matthias Clasen , Thu, 7 Oct 2010 07:55:42 -0400: > >> On Thu, Oct 7, 2010 at 7:51 AM, Vincent Untz wrote: >> >> > I think the right solution is to improve l10n.gnome.org so that it can >> > interacts with other platforms (like transifex) where there is a team >> > control. >> > >> > Alternatively we can also investigate something similar to where >> > transifex is heading to: host the po files on l10n.gnome.org and add >> > some autofoo magic that would download the po files from there during >> > 'make dist'. >> >> I think moving toward transifex or a similar distributed approach for >> translations is the way to go. > > When it comes to consistency in translation (which matters), not > distributed approach, but centralized effort of community is the key to > success. That's what I learned from several years of translating FLOSS > projects. And I guess that other translators observed that as well. I agree. From the point of a translation coordinator. The best choise is to have all translations in one place that he/she controls. Alternatively, when that fails the second best options is to have all the translations centered in as few solutions as possible. And there are important things to look for when determining which to use, but I'll dive into that in the separate thread that has been stated. \Kenneth ___ gnome-i18n mailing list gnome-i18n@gnome.org http://mail.gnome.org/mailman/listinfo/gnome-i18n
Re: New module proposal: Clutter core
Hi! Matthias Clasen , Thu, 7 Oct 2010 07:55:42 -0400: > On Thu, Oct 7, 2010 at 7:51 AM, Vincent Untz wrote: > > > I think the right solution is to improve l10n.gnome.org so that it can > > interacts with other platforms (like transifex) where there is a team > > control. > > > > Alternatively we can also investigate something similar to where > > transifex is heading to: host the po files on l10n.gnome.org and add > > some autofoo magic that would download the po files from there during > > 'make dist'. > > I think moving toward transifex or a similar distributed approach for > translations is the way to go. When it comes to consistency in translation (which matters), not distributed approach, but centralized effort of community is the key to success. That's what I learned from several years of translating FLOSS projects. And I guess that other translators observed that as well. My 10 hellers though, Petr Kovar ___ gnome-i18n mailing list gnome-i18n@gnome.org http://mail.gnome.org/mailman/listinfo/gnome-i18n
Re: New module proposal: Clutter core
On Thu, Oct 7, 2010 at 7:51 AM, Vincent Untz wrote: > I think the right solution is to improve l10n.gnome.org so that it can > interacts with other platforms (like transifex) where there is a team > control. > > Alternatively we can also investigate something similar to where > transifex is heading to: host the po files on l10n.gnome.org and add > some autofoo magic that would download the po files from there during > 'make dist'. I think moving toward transifex or a similar distributed approach for translations is the way to go. ___ gnome-i18n mailing list gnome-i18n@gnome.org http://mail.gnome.org/mailman/listinfo/gnome-i18n
Re: New module proposal: Clutter core
Le mardi 05 octobre 2010, à 18:41 +0200, Johannes Schmid a écrit : > Hi! > > > I'm pretty sure the GNOME infrastructure could do the same thing: get > > the POT file from git.clutter-project.org (it's generated by gettext and > > stored in the repository anyway); send me an email with the PO file once > > the coordinator has reviewed the contribution. I could even allow commit > > access to a branch, or a user repository that I can pull from. > > That would work (in might be a good idea to implement for git > repositories in general). We didn't manage to commit to GNOME git yet > though and I doubt it would be easier on other repositories. It's > something that could be considered in long-term, also for some > freedesktop modules. > > > alternatively, GNOME could have a private Clutter core repository for > > i18n purposes alone - after all, we're using Git. > > We do that for some other modules and it works OK. > > Note: As stated before I am not so much concerned about clutter itself > but about our general policy on handling external repositories. For git > it might kind of work but for other things like launchpad it could > become worse. I think the right solution is to improve l10n.gnome.org so that it can interacts with other platforms (like transifex) where there is a team control. Alternatively we can also investigate something similar to where transifex is heading to: host the po files on l10n.gnome.org and add some autofoo magic that would download the po files from there during 'make dist'. Vincent -- Les gens heureux ne sont pas pressés. ___ gnome-i18n mailing list gnome-i18n@gnome.org http://mail.gnome.org/mailman/listinfo/gnome-i18n
Re: New module proposal: Clutter core
On Tue, 2010-10-05 at 22:47 +0200, Claude Paroz wrote: > Le mardi 05 octobre 2010 à 18:41 +0200, Johannes Schmid a écrit : > > Hi! > > > > > I'm pretty sure the GNOME infrastructure could do the same thing: get > > > the POT file from git.clutter-project.org (it's generated by gettext and > > > stored in the repository anyway); send me an email with the PO file once > > > the coordinator has reviewed the contribution. I could even allow commit > > > access to a branch, or a user repository that I can pull from. > > > > That would work (in might be a good idea to implement for git > > repositories in general). We didn't manage to commit to GNOME git yet > > though and I doubt it would be easier on other repositories. > > The blocker is not technical, but more about the policy to accept > commits from an application (and create security hooks). I know it's on > Christer's TODO so there is some hope :-) > > If clutter is open to give D-L a commit access, we could try to set up > something in the not so long-term... sure; as I said, I can create a separate repository and then periodically sync the main repo to it. where this repository should live I can leave it entirely up to the i18n team; I can create a user repo on clutter-project.org and accept coordinators keys - or it can live on gnome.org, so the auth infrastructure would already be in place. ciao, Emmanuele. -- W: http://www.emmanuelebassi.name B: http://blogs.gnome.org/ebassi ___ gnome-i18n mailing list gnome-i18n@gnome.org http://mail.gnome.org/mailman/listinfo/gnome-i18n
Re: New module proposal: Clutter core
Le mardi 05 octobre 2010 à 18:41 +0200, Johannes Schmid a écrit : > Hi! > > > I'm pretty sure the GNOME infrastructure could do the same thing: get > > the POT file from git.clutter-project.org (it's generated by gettext and > > stored in the repository anyway); send me an email with the PO file once > > the coordinator has reviewed the contribution. I could even allow commit > > access to a branch, or a user repository that I can pull from. > > That would work (in might be a good idea to implement for git > repositories in general). We didn't manage to commit to GNOME git yet > though and I doubt it would be easier on other repositories. The blocker is not technical, but more about the policy to accept commits from an application (and create security hooks). I know it's on Christer's TODO so there is some hope :-) If clutter is open to give D-L a commit access, we could try to set up something in the not so long-term... Cheers, Claude > It's > something that could be considered in long-term, also for some > freedesktop modules. > > > alternatively, GNOME could have a private Clutter core repository for > > i18n purposes alone - after all, we're using Git. > > We do that for some other modules and it works OK. > > Note: As stated before I am not so much concerned about clutter itself > but about our general policy on handling external repositories. For git > it might kind of work but for other things like launchpad it could > become worse. > > Regards, > Johannes ___ gnome-i18n mailing list gnome-i18n@gnome.org http://mail.gnome.org/mailman/listinfo/gnome-i18n
Re: New module proposal: Clutter core
Hi! > I'm pretty sure the GNOME infrastructure could do the same thing: get > the POT file from git.clutter-project.org (it's generated by gettext and > stored in the repository anyway); send me an email with the PO file once > the coordinator has reviewed the contribution. I could even allow commit > access to a branch, or a user repository that I can pull from. That would work (in might be a good idea to implement for git repositories in general). We didn't manage to commit to GNOME git yet though and I doubt it would be easier on other repositories. It's something that could be considered in long-term, also for some freedesktop modules. > alternatively, GNOME could have a private Clutter core repository for > i18n purposes alone - after all, we're using Git. We do that for some other modules and it works OK. Note: As stated before I am not so much concerned about clutter itself but about our general policy on handling external repositories. For git it might kind of work but for other things like launchpad it could become worse. Regards, Johannes signature.asc Description: This is a digitally signed message part ___ gnome-i18n mailing list gnome-i18n@gnome.org http://mail.gnome.org/mailman/listinfo/gnome-i18n
Re: New module proposal: Clutter core
On Tue, 2010-10-05 at 10:25 +0200, Johannes Schmid wrote: > >> > I want to raise the point again, that the separate git server is > >> painful > >> > for translators which is the main reason that I dislike it. (see > >> > http://mail.gnome.org/archives/gnome-i18n/2010-July/msg00075.html and > >> > follow ups) > >> I agree with Johannes, especially about the quality. As an easy fix > >> for this, couldn't we just keep the translations in a git.gnome.org > >> module? It would not allow us to run intltool-udpate and all that, but > >> that would probably be ok as long and the maintainers would fetch new > >> translations and update translation files with new strings regularly. > > > > Oh, so here we go again. Thanks for raising this issue, guys. I'm, too, > > convinced that it's important for the future of the GTP and GNOME > > translation teams to decide what should we require from core GNOME modules > > (or however we label/define it). > Moving this discussion back to desktop-devel-list where it should have > stayed with a CC'd gnome-i18n. thanks for the feedback. I'm not even going to try and convince the i18n teams - mostly because I agree: the GNOME i18n teams should be using a single infrastructure and not 20 different ones. now, how do we go from here to there is probably worth discussing. I cannot move Clutter to gnome.org; it's simply unfeasible for various reasons, one of which is that the Clutter Project is not just used by GNOME. this is similar to GStreamer, or Cairo, which are hosted on freedesktop.org. another thing to consider is that translating Clutter is probably never going to be a priority: the messages are mostly going to be errors; there are no widgets, complex or otherwise composed by user-facing text; and the only other translatable, user-facing strings are property nicks and blurbs that can only be visible if a UI builder tool is introspecting them to create a UI. currently, Clutter uses Transifex in a fairly passive way: I get emails for new PO files, I copy them into the repo and commit them. I checked if Transifex has a way to handle custom repositories, so that I could allow direct commit access to a branch, and then periodically merge the branch back into master; it doesn't seem to be possible without getting the Transifex admins to handle custom repositories - and, honestly, right now the burden is not at all high. I'm pretty sure the GNOME infrastructure could do the same thing: get the POT file from git.clutter-project.org (it's generated by gettext and stored in the repository anyway); send me an email with the PO file once the coordinator has reviewed the contribution. I could even allow commit access to a branch, or a user repository that I can pull from. alternatively, GNOME could have a private Clutter core repository for i18n purposes alone - after all, we're using Git. ciao, Emmanuele. -- W: http://www.emmanuelebassi.name B: http://blogs.gnome.org/ebassi ___ gnome-i18n mailing list gnome-i18n@gnome.org http://mail.gnome.org/mailman/listinfo/gnome-i18n
Re: New module proposal: Clutter core
Hi! > Kenneth Nielsen , Mon, 4 Oct 2010 15:45:02 +0200: > >> 2010/10/4 Johannes Schmid : >> > Hi! >> > >> >> Clutter is still hosted on a separate server because the Clutter >> >> Project wants to be an umbrella for a set of projects, like language >> >> bindings, toolkits, and applications that may or may not be related >> to >> >> the GNOME Project. we're fairly liberal with giving people access to >> >> the repository, and we have infrastructure in place for user >> >> repositories for contributors. the Bugzilla instance is still in >> place >> >> because Clutter is used in non-GNOME projects that might need >> >> restricted access. >> > >> > I want to raise the point again, that the separate git server is >> painful >> > for translators which is the main reason that I dislike it. (see >> > http://mail.gnome.org/archives/gnome-i18n/2010-July/msg00075.html and >> > follow ups) >> > >> > Basically the point is that if we allow core modules to be hosted >> > elsewhere we can shut down the GNOME Translation Project as it exists >> > now completely because our whole quality work with coordinators and >> > reviewers will become obsolete. GNOME has a very long and good >> > tradition of high-level and consistent translations which would get >> > lost. >> > >> > The point is not that important for clutter which probably doesn't >> > contain many user visible strings but if we they yes here it will be >> > difficult to say no with other modules. >> > >> > Needless to say that I of course in general like the idea of having >> > clutter as a core module. >> > >> > Regards, >> > Johannes >> >> I agree with Johannes, especially about the quality. As an easy fix >> for this, couldn't we just keep the translations in a git.gnome.org >> module? It would not allow us to run intltool-udpate and all that, but >> that would probably be ok as long and the maintainers would fetch new >> translations and update translation files with new strings regularly. > > Oh, so here we go again. Thanks for raising this issue, guys. I'm, too, > convinced that it's important for the future of the GTP and GNOME > translation teams to decide what should we require from core GNOME modules > (or however we label/define it). > > I'm not really sure whether it's realistic to expect any [code hosting] > infrastructure movement in this case, but as Aron Xu suggested the last > time > this discussion came up, we could try to propose doing the clutter l10n > management the system-tools-backends way, i.e. maintaining a Git clone on > git.gnome.org. > > Certainly, clutter is a fairly different piece of software, and this > cloning > workflow may have some clear shortcomings. E.g. maintenance burden for > developers, depending on developers' time, no guarantee of up-to-date POT > files, as our translators are used to and expect it esp. when dealing with > tight deadlines, that is every six months. > > If nothing more, we could at least persuade developers to stick to the > more > closely managed module l10n community (if such a word is appropriate > here), > so to not allow anyone on the net to submit translation work with, from > the > GTP perspective, varying quality. Moving this discussion back to desktop-devel-list where it should have stayed with a CC'd gnome-i18n. Regards, Johannes ___ gnome-i18n mailing list gnome-i18n@gnome.org http://mail.gnome.org/mailman/listinfo/gnome-i18n
Re: New module proposal: Clutter core
Hi! Kenneth Nielsen , Mon, 4 Oct 2010 15:45:02 +0200: > 2010/10/4 Johannes Schmid : > > Hi! > > > >> Clutter is still hosted on a separate server because the Clutter > >> Project wants to be an umbrella for a set of projects, like language > >> bindings, toolkits, and applications that may or may not be related to > >> the GNOME Project. we're fairly liberal with giving people access to > >> the repository, and we have infrastructure in place for user > >> repositories for contributors. the Bugzilla instance is still in place > >> because Clutter is used in non-GNOME projects that might need > >> restricted access. > > > > I want to raise the point again, that the separate git server is painful > > for translators which is the main reason that I dislike it. (see > > http://mail.gnome.org/archives/gnome-i18n/2010-July/msg00075.html and > > follow ups) > > > > Basically the point is that if we allow core modules to be hosted > > elsewhere we can shut down the GNOME Translation Project as it exists > > now completely because our whole quality work with coordinators and > > reviewers will become obsolete. GNOME has a very long and good > > tradition of high-level and consistent translations which would get > > lost. > > > > The point is not that important for clutter which probably doesn't > > contain many user visible strings but if we they yes here it will be > > difficult to say no with other modules. > > > > Needless to say that I of course in general like the idea of having > > clutter as a core module. > > > > Regards, > > Johannes > > I agree with Johannes, especially about the quality. As an easy fix > for this, couldn't we just keep the translations in a git.gnome.org > module? It would not allow us to run intltool-udpate and all that, but > that would probably be ok as long and the maintainers would fetch new > translations and update translation files with new strings regularly. Oh, so here we go again. Thanks for raising this issue, guys. I'm, too, convinced that it's important for the future of the GTP and GNOME translation teams to decide what should we require from core GNOME modules (or however we label/define it). I'm not really sure whether it's realistic to expect any [code hosting] infrastructure movement in this case, but as Aron Xu suggested the last time this discussion came up, we could try to propose doing the clutter l10n management the system-tools-backends way, i.e. maintaining a Git clone on git.gnome.org. Certainly, clutter is a fairly different piece of software, and this cloning workflow may have some clear shortcomings. E.g. maintenance burden for developers, depending on developers' time, no guarantee of up-to-date POT files, as our translators are used to and expect it esp. when dealing with tight deadlines, that is every six months. If nothing more, we could at least persuade developers to stick to the more closely managed module l10n community (if such a word is appropriate here), so to not allow anyone on the net to submit translation work with, from the GTP perspective, varying quality. Best, Petr Kovar ___ gnome-i18n mailing list gnome-i18n@gnome.org http://mail.gnome.org/mailman/listinfo/gnome-i18n
Re: New module proposal: Clutter core
2010/10/4 Johannes Schmid : > Hi! > >> Clutter is still hosted on a separate server because the Clutter Project >> wants to be an umbrella for a set of projects, like language bindings, >> toolkits, and applications that may or may not be related to the GNOME >> Project. we're fairly liberal with giving people access to the >> repository, and we have infrastructure in place for user repositories >> for contributors. the Bugzilla instance is still in place because >> Clutter is used in non-GNOME projects that might need restricted access. > > I want to raise the point again, that the separate git server is painful > for translators which is the main reason that I dislike it. (see > http://mail.gnome.org/archives/gnome-i18n/2010-July/msg00075.html and > follow ups) > > Basically the point is that if we allow core modules to be hosted > elsewhere we can shut down the GNOME Translation Project as it exists now > completely because our whole quality work with coordinators and reviewers > will become obsolete. GNOME has a very long and good tradition of > high-level and consistent translations which would get lost. > > The point is not that important for clutter which probably doesn't contain > many user visible strings but if we they yes here it will be difficult to > say no with other modules. > > Needless to say that I of course in general like the idea of having > clutter as a core module. > > Regards, > Johannes I agree with Johannes, especially about the quality. As an easy fix for this, couldn't we just keep the translations in a git.gnome.org module? It would not allow us to run intltool-udpate and all that, but that would probably be ok as long and the maintainers would fetch new translations and update translation files with new strings regularly. Regards Kenneth ___ gnome-i18n mailing list gnome-i18n@gnome.org http://mail.gnome.org/mailman/listinfo/gnome-i18n