Hi All,
Let's dig deep into it and I am sure we will find the way to do this much better. Please don't blame anyone just focus on the issues and their possible solutions. I find to following issues and please help if you have others:
- the translations not appears in the factory from svn and we don't know the process - it looks the deadline is not well communicated (we need more alerts) about deadlines and about - ???
idea 1: I have a script that checks svn website every 5 or 10 minutes and send me a mail if something had changed in the 50-pot directories. If you need I can forward it the list or we can create a separate list or google groups for that kind messages. However opensuse already have process on the commit list (I think).
idea 2: We have to analyze the factory src and we compare with the svn automatically. We need a warning if translation not go into the factory within 5 days.
idea 3: We need more warning mails before deadlines. One status mail on every week and one mail per day right before the deadline. The translations without any commits around in the deadline should be warn directly. (I like that at Mozilla. They contact me directly how can they help me to complete the translation and even both of us know they could not help me, this is a nice warning that I am not follow the schedule and maybe they have to find another volunteer)
idea 4 We have to check the merge process. If the total message numbers not the same for all languages. We have to find out what is wrong until the total numbers equals to the pot's total number
idea 5 We have to build l10n volunteers respect and community. I am not sure why we should deliver languages where translations are under 75%. (at Mozilla this is 100%). We have to contact to the translators and find out what happens and maybe we have find more/new volunteers.
Let me know your ideas and thoughts about mine.
thanks kalman
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256
Kálmán Kéménczy escribió:
idea 1: I have a script that checks svn website every 5 or 10 minutes and send me a mail if something had changed in the 50-pot directories. If you need I can forward it the list or we can create a separate list or google groups for that kind messages. However opensuse already have process on the commit list (I think).
Regarding this, some teams are using vertaal (www.vertaal.com.ar) which already have this feature. I don't think opensuse have a hook for this implemented.
idea 3: We need more warning mails before deadlines. One status mail on every week and one mail per day right before the deadline. The translations without any commits around in the deadline should be warn directly. (I like that at Mozilla. They contact me directly how can they help me to complete the translation and even both of us know they could not help me, this is a nice warning that I am not follow the schedule and maybe they have to find another volunteer)
Seems good.
idea 4 We have to check the merge process. If the total message numbers not the same for all languages. We have to find out what is wrong until the total numbers equals to the pot's total number
Vertaal do that too :-)
thanks kalman
- -- Kind regards.
Regarding this, some teams are using vertaal (www.vertaal.com.ar) which already have this feature.
let me try. anyone has experience with transifex? I would like to try that too.
k
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256
Kálmán Kéménczy escribió:
Regarding this, some teams are using vertaal (www.vertaal.com.ar) which already have this feature.
let me try. anyone has experience with transifex? I would like to try that too.
I tried transifex, didn't like the stable version and the new one was very young at the time oS 11.2 translations started. So I wrote Vertaal to accomplish this task, which IHMO is quite stable right now. Though is similar to transifex it has some features built thinking on openSUSE (though it could work on any project).
- -- Kind regards.
I tried transifex, didn't like the stable version and the new one was very young at the time oS 11.2 translations started. So I wrote Vertaal to accomplish this task, which IHMO is quite stable right now. Though is similar to transifex it has some features built thinking on openSUSE (though it could work on any project).
I am getting exited! I send a request for Hungarian part. Please let me know if I could start to work with it. Maybe we could talk about it private. thanks kalman
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Content-ID: alpine.LSU.2.00.0911080420300.5098@nimrodel.valinor
On Saturday, 2009-11-07 at 18:02 +0100, Kálmán Kéménczy wrote:
Hi All,
Let's dig deep into it and I am sure we will find the way to do this much better. Please don't blame anyone just focus on the issues and their possible solutions. I find to following issues and please help if you have others:
Ok.
- the translations not appears in the factory from svn and we don't
know the process
- it looks the deadline is not well communicated (we need more alerts)
about deadlines and about
- ???
idea 1: I have a script that checks svn website every 5 or 10 minutes and send me a mail if something had changed in the 50-pot directories. If you need I can forward it the list or we can create a separate list or google groups for that kind messages. However opensuse already have process on the commit list (I think).
Those teams using "vertaal" get this warning on the email of the person that has taken responsibility of a particular file.
idea 2: We have to analyze the factory src and we compare with the svn automatically. We need a warning if translation not go into the factory within 5 days.
Most certainly.
IMO, each dev should have a process, part of the makefile perhaps, that pulls all the translations from the svn before "making" the project.
Or at least, warn the dev automatically that his project "po" directory is not in sync with our svn, and perhaps even block the submission of the rpm till solved or forced.
idea 3: We need more warning mails before deadlines. One status mail on every week and one mail per day right before the deadline. The translations without any commits around in the deadline should be warn directly. (I like that at Mozilla. They contact me directly how can they help me to complete the translation and even both of us know they could not help me, this is a nice warning that I am not follow the schedule and maybe they have to find another volunteer)
Certainly. And they get mozilla translated, a very nice impression.
idea 4 We have to check the merge process. If the total message numbers not the same for all languages. We have to find out what is wrong until the total numbers equals to the pot's total number
Not only number of messages, but whether the messages match (if they don't, they are not displayed in our language, but it reverts to English).
idea 5 We have to build l10n volunteers respect and community. I am not sure why we should deliver languages where translations are under 75%. (at Mozilla this is 100%). We have to contact to the translators and find out what happens and maybe we have find more/new volunteers.
Most certainly.
- -- Cheers, Carlos E. R.
idea 5 We have to build l10n volunteers respect and community. I am not sure why we should deliver languages where translations are under 75%. (at Mozilla this is 100%). We have to contact to the translators and find out what happens and maybe we have find more/new volunteers.
Hi,
For idea 5 I think that opensuse does something like that, at least for yast. There is a package yast2-trans-stats with the description "The package contains statistic files, one file per language. With the help of these statistics we can warn you if you select a language for installation which is unsufficiently translated." which I think did this but I don't know how it works.
Regards,
Luiz
2009/11/7 Kálmán Kéménczy kkemenczy@opensuse.org
idea 5 We have to build l10n volunteers respect and community. I am not sure why we should deliver languages where translations are under 75%. (at Mozilla this is 100%). We have to contact to the translators and find out what happens and maybe we have find more/new volunteers.
That's not very respectfull for small or minority languages that strive to have the most translated they can, dealing with openSUSE, Gnome, KDE, Mozilla... Try to find new volounteers does not always succeed. So what? Throw the language away? What's the point to have Yast 100% translated (even some very specific and technician parts) if half of the desktop is untranslated?
On KDE, the language is released if the essentials are met: desktop_kdelibs.po : 75% desktop_l10n.po : 75% kdebase : 75% kdelibs4.po : 90%
These files/folders are about 20% of the whole KDE project but they are the core. The other parts are not evaluated for release.
On GNOME, they don't care.
According to the stats, following your idea would make 23 languages only. There are 60 languages translated on openSUSE. http://i18n.opensuse.org/stats/trunk/toplist.php
Speaking of my case, I'm very proud to have managed to come from 45% to 57% from 11.1 to 11.2 alone, despite I try to find other translators. And I work also on KDE and Firefox (with a very few other people). Should I give up or spend all my hometime translating it? Or stop translating KDE to be sure openSUSE gets 75%?
Mozilla is far smaller in number of strings than openSUSE. Translations on openSUSE are far more work intensive than, for example, Mozilla or KDE want it's more technical and it includes the documentation in the GUI. They're several POT with weird name with not any explanation of what it's for on the web. Should I translate it badly to get the %age or should I leave it now untranslated?
I can understand than some languages around the 20% or so are a bit "short" but for the others, please think about the consequences. If an "essential" requirement is set, it must be defined wisely and the POT files have to be cleaned up from some outdated pot's (kinternet for example...).
Regards,
Jean Cayron wa-team
Let me know your ideas and thoughts about mine.
thanks kalman -- To unsubscribe, e-mail: opensuse-translation+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-translation+help@opensuse.org
Jean Cayron jean.cayron@gmail.com writes:
2009/11/7 Kálmán Kéménczy kkemenczy@opensuse.org
idea 5 We have to build l10n volunteers respect and community. I am not sure why we should deliver languages where translations are under 75%. (at Mozilla this is 100%). We have to contact to the translators and find out what happens and maybe we have find more/new volunteers.
That's not very respectfull for small or minority languages that strive to have the most translated they can, dealing with openSUSE, Gnome, KDE, Mozilla... Try to find new volounteers does not always succeed. So what? Throw the language away? What's the point to have Yast 100% translated (even some very specific and technician parts) if half of the desktop is untranslated?
In YaST, there used to be a warning telling the user that language X is not complete, if he selects it for installation. IIRC, this warning pops up at 70% or 80%. I think we should improve this mechanism. Yes, you are right, for the regular user completely translated applications are more useful then a complete translated admin tool such as YaST.
Even if a language is very incompletely we should publish it and telling the user the status as detailed as possible. The only reason not to distribute or enable a language is, when the quality of the translation is dubious.
Hi Jean,
On Tue, Nov 10, 2009 at 1:24 PM, Jean Cayron jean.cayron@gmail.com wrote:
2009/11/7 Kálmán Kéménczy kkemenczy@opensuse.org
idea 5 We have to build l10n volunteers respect and community. I am not sure why we should deliver languages where translations are under 75%. (at Mozilla this is 100%). We have to contact to the translators and find out what happens and maybe we have find more/new volunteers.
That's not very respectfull for small or minority languages that strive to have the most translated they can, dealing with openSUSE, Gnome, KDE, Mozilla...
Try to find new volounteers does not always succeed. So what? Throw the language away? What's the point to have Yast 100% translated (even some very specific and technician parts) if half of the desktop is untranslated?
First of all, thanks for your feedback. I am just talking about ideas here. There are no decisions (and I am not the one who could or want decide anything alone).
About the translation percent. I am not talking about YaST, I am talking about the distribution we release. And now I see I was not clear. The localization is the hardest part when we are talking about a whole distribution. How can I say to a school, government or users that openSUSE is localized, when it's not. And we have no idea at all the l10n status of any release. We really don't know what is missing or what we have. The stat page is represent 5% (or less) of the distribution and this is not good overview.
I think we have to find out when we could say something is localized. Which components are key and where should we improve. Maybe we have to find new contributors on different languages where there was no commit for years.
On KDE, the language is released if the essentials are met: desktop_kdelibs.po : 75% desktop_l10n.po : 75% kdebase : 75% kdelibs4.po : 90%
These files/folders are about 20% of the whole KDE project but they are the core. The other parts are not evaluated for release.
On GNOME, they don't care.
According to the stats, following your idea would make 23 languages only. There are 60 languages translated on openSUSE. http://i18n.opensuse.org/stats/trunk/toplist.php
Speaking of my case, I'm very proud to have managed to come from 45% to 57% from 11.1 to 11.2 alone, despite I try to find other translators.
We should and we are proud of this effort as well.
And I work also on KDE and Firefox (with a very few other people). Should I give up or spend all my hometime translating it? Or stop translating KDE to be sure openSUSE gets 75%?
No, we should build community for each languages. The one man show not works well in the community development.
Mozilla is far smaller in number of strings than openSUSE. Translations on openSUSE are far more work intensive than, for example, Mozilla or KDE want it's more technical and it includes the documentation in the GUI. They're several POT with weird name with not any explanation of what it's for on the web. Should I translate it badly to get the %age or should I leave it now untranslated?
imho, you are already know the answer for that. Bad translations are worst than no translations.
I can understand than some languages around the 20% or so are a bit "short" but for the others, please think about the consequences. If an "essential" requirement is set, it must be defined wisely and the POT files have to be cleaned up from some outdated pot's (kinternet for example...).
We are here to talk about it. And again, I am just one small piece of the list and I will not and can't judge anything. I just would like a conversations between us, how can we leverage the processes.
thanks kalman
Hi everyone,
We're of course discussing and it's only ideas.
2009/11/10 Kálmán Kéménczy kkemenczy@opensuse.org:
Hi Jean,
On Tue, Nov 10, 2009 at 1:24 PM, Jean Cayron jean.cayron@gmail.com wrote:
2009/11/7 Kálmán Kéménczy kkemenczy@opensuse.org
idea 5 We have to build l10n volunteers respect and community. I am not sure why we should deliver languages where translations are under 75%. (at Mozilla this is 100%). We have to contact to the translators and find out what happens and maybe we have find more/new volunteers.
That's not very respectfull for small or minority languages that strive to have the most translated they can, dealing with openSUSE, Gnome, KDE, Mozilla...
Try to find new volounteers does not always succeed. So what? Throw the language away? What's the point to have Yast 100% translated (even some very specific and technician parts) if half of the desktop is untranslated?
First of all, thanks for your feedback. I am just talking about ideas here. There are no decisions (and I am not the one who could or want decide anything alone).
About the translation percent. I am not talking about YaST, I am talking about the distribution we release. And now I see I was not clear. The localization is the hardest part when we are talking about a whole distribution. How can I say to a school, government or users that openSUSE is localized, when it's not. And we have no idea at all the l10n status of any release. We really don't know what is missing or what we have. The stat page is represent 5% (or less) of the distribution and this is not good overview.
In that case, the "Warning" made about Yast could be used for the whole thing but we should keep "developping" languages. It should be clear in the promotion (wiki, sale DVD...) what are the "full experience localised" languages. But there, the problem is how to "rate" it. Rating on a % of KDE, Gnome and openSUSE has no sens, as I said, want there are loads of small programs not widely used on both desktops. Essentials for KDE makes more sense. But again, openSUSE doesn't ship unofficial translations of KDE, so it's already the case. Other key applications are OpenOffice and Firefox. For these we can say: OK it's officially localised. For Gnome I don't have a clue of how to deal with it.
Or it can be done on the packages shipped on the OSS repo.
And what about GNU tools ;-)
Idea: Yast: X % yes/no lcn: X % yes/no Gnome: ???? yes/no KDE: official KDE released language yes/no Firefox: official Firefox released language y/n OpenOffice: official released language y/n other important projects, application?
How to make stats about all these projects? That's a lot of work to do for all languages, indeed and I'm not sure that scripty can do it.
And there we don't speek about quality, only abou %age. I remembered having heard a sad story about a Mandriva based distro "Caxa Magica" or something similar that had a contract with Portuguese government for school computers. In their distro there was GCompris and it was translated in Portuguese by one of his creator, from Portuguese origin but living in France. Some translations were a bit "unusual" and the teachers made a fuzz about the "badly translated educational program", the press got involved and so on, despite the GCompris community brought a quality update on that translations within the week. It ended up a bit sadly with the educational minister explaining how to remove the "evil" application without mentioning that there was an update available...
There is no good way, except feed-back from a release to another, to rate that.
But again, if something is done, I'm for the warning, not the withdraw from the translations.
I think we have to find out when we could say something is localized. Which components are key and where should we improve. Maybe we have to find new contributors on different languages where there was no commit for years.
On KDE, the language is released if the essentials are met: desktop_kdelibs.po : 75% desktop_l10n.po : 75% kdebase : 75% kdelibs4.po : 90%
These files/folders are about 20% of the whole KDE project but they are the core. The other parts are not evaluated for release.
On GNOME, they don't care.
According to the stats, following your idea would make 23 languages only. There are 60 languages translated on openSUSE. http://i18n.opensuse.org/stats/trunk/toplist.php
Speaking of my case, I'm very proud to have managed to come from 45% to 57% from 11.1 to 11.2 alone, despite I try to find other translators.
We should and we are proud of this effort as well.
And I work also on KDE and Firefox (with a very few other people). Should I give up or spend all my hometime translating it? Or stop translating KDE to be sure openSUSE gets 75%?
No, we should build community for each languages. The one man show not works well in the community development.
I know but more you have speakers, more you have volunteers. And more you have systems/apps localised, more you attract people. It's a bit the case for me now, some people start to propose their help because they've been attracted with a product.
Mozilla is far smaller in number of strings than openSUSE. Translations on openSUSE are far more work intensive than, for example, Mozilla or KDE want it's more technical and it includes the documentation in the GUI. They're several POT with weird name with not any explanation of what it's for on the web. Should I translate it badly to get the %age or should I leave it now untranslated?
imho, you are already know the answer for that. Bad translations are worst than no translations.
I can understand than some languages around the 20% or so are a bit "short" but for the others, please think about the consequences. If an "essential" requirement is set, it must be defined wisely and the POT files have to be cleaned up from some outdated pot's (kinternet for example...).
We are here to talk about it. And again, I am just one small piece of the list and I will not and can't judge anything. I just would like a conversations between us, how can we leverage the processes.
thanks kalman
No personal problem with you Kalman, if you pass and visit me in Belgium, I'll offer you a free homemade beer with pleasure ;-).
Regards,
Jean
Rating on a % of KDE, Gnome and openSUSE has no sens, as I said, want there are loads of small programs not widely used on both desktops.
Somehow we have to collect the impact rate of translations. No idea how.
Essentials for KDE makes more sense. But again, openSUSE doesn't ship unofficial translations of KDE, so it's already the case. Other key applications are OpenOffice and Firefox. For these we can say: OK it's officially localised. For Gnome I don't have a clue of how to deal with it.
They release the list of the supported languages (base gui 80%+) in the release notes http://library.gnome.org/misc/release-notes/2.28/
Or it can be done on the packages shipped on the OSS repo.
And what about GNU tools ;-)
it's more important than it looks. This is in the deep, but lots of program has been built on it.
Idea: Yast: X % yes/no
we should find the important modules (eg. install and modules with default installation is a must, s390 not really counts)
How can we get statistics about module usage from the openSUSE users?
lcn: X % yes/no
same as yast
Gnome: ???? yes/no
there is official yes/no
KDE: official KDE released language yes/no Firefox: official Firefox released language y/n OpenOffice: official released language y/n other important projects, application?
need a survey or we have to judge it by ourself
How to make stats about all these projects? That's a lot of work to do for all languages, indeed and I'm not sure that scripty can do it.
Well I have been working on it for years. And I really hope I could show something. This script is analyze the factory source itself. But we will see....
And there we don't speek about quality, only abou %age. I remembered
You cannot measure the quality easily, however we can check few things: consistency.
for example, how we translate Cancel? http://en.hu.open-tran.eu/suggest/cancel not consistent :( we have to improve it, because for the users this is one system not small pieces on each others.
There is no good way, except feed-back from a release to another, to rate that.
But again, if something is done, I'm for the warning, not the withdraw from the translations.
agree
No personal problem with you Kalman, if you pass and visit me in Belgium, I'll offer you a free homemade beer with pleasure ;-).
thanks, I can't wait :)
best, kalman
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On Wednesday, 2009-11-11 at 08:26 +0100, Jean Cayron wrote:
...
And what about GNU tools ;-)
Badly. :-(
I had a look at it once, and I didn't like the requirements: they want translators to sign a paper before accepting translations. No way, thanks.
And that without mentioning lack of good tools to translate, for example, man pages... How can you tell some one to RTFM (flipping :-p ) if there is no translation?
- -- Cheers, Carlos E. R.
translation@lists.opensuse.org
-
Carlos E. R. -
Carlos E. R. -
Gabriel -
Jean Cayron -
Karl Eichwalder -
Kálmán Kéménczy -
Luiz Fernando Ranghetti