Hi everyone!
I want to translate Etoys (I saw it's already translated in Spanish, Japanese etc.) in Romanian, but can't find the .mo files with the existing translations.
Can anybody point me these .mo files' location and maybe the steps I have to take in order my translation to appear in the official release of Etoys?
Thanks in advance. tb
On Jul 21, 2007, at 7:06 , Tibi wrote:
Hi everyone!
I want to translate Etoys (I saw it's already translated in Spanish, Japanese etc.) in Romanian,
Great, Tibi!
but can't find the .mo files with the existing translations.
Can anybody point me these .mo files' location and maybe the steps I have to take in order my translation to appear in the official release of Etoys?
Etoys used a translation system different from gettext until now. All the translations that are in there use Squeak's own translation mechanism. You could use that - press Alt-Shift-W to bring up a menu and select open... - Language editor.
However, we are working on switching to gettext to make it simpler for others to contribute translations:
http://dev.laptop.org/ticket/2133
We already can generate .pot files, but are not completely satisfied with the results yet:
On Jul 12, 2007, at 17:39 , Takashi Yamamiya wrote:
You can extract .pot files with each class category from OLPC etoys image with
GetTextExporter2 new export.
Although, there are some issues, apparently you might notice that sometimes a keyword shows twice, it looks good start.
Takashi, I just notice that all the tile commands are missing, perhaps because these are no literal senders of #translated?
Tibi, I'd ask you to wait a bit on the translations. If you want to start now, use the old tools, or work off the .pot files I attached, but be prepared that these will change soon (they will be merged to fewer files).
Thank you!
- Bert -
Hi Bert!
Thank you for responding.
On 7/21/07, Bert Freudenberg bert@freudenbergs.de wrote:
On Jul 21, 2007, at 7:06 , Tibi wrote:
Hi everyone!
I want to translate Etoys (I saw it's already translated in Spanish, Japanese etc.) in Romanian,
Great, Tibi!
but can't find the .mo files with the existing translations.
Can anybody point me these .mo files' location and maybe the steps I have to take in order my translation to appear in the official release of Etoys?
Etoys used a translation system different from gettext until now. All the translations that are in there use Squeak's own translation mechanism. You could use that - press Alt-Shift-W to bring up a menu and select open... - Language editor.
However, we are working on switching to gettext to make it simpler for others to contribute translations:
http://dev.laptop.org/ticket/2133We already can generate .pot files, but are not completely satisfied with the results yet:
On Jul 12, 2007, at 17:39 , Takashi Yamamiya wrote:
You can extract .pot files with each class category from OLPC etoys image with
GetTextExporter2 new export.
Although, there are some issues, apparently you might notice that sometimes a keyword shows twice, it looks good start.
Takashi, I just notice that all the tile commands are missing, perhaps because these are no literal senders of #translated?
Tibi, I'd ask you to wait a bit on the translations. If you want to start now, use the old tools, or work off the .pot files I attached, but be prepared that these will change soon (they will be merged to fewer files).
Could you estimate how soon the changes will come?
Thank you!
- Bert -
I thank you too. tb
Hi Bert,
Bert Freudenberg wrote:
On Jul 22, 2007, at 9:01 , Tibi wrote:
Could you estimate how soon the changes will come?
Takashi would know better, but I'd hope very soon, a few days only. It's a high-priority task.
O.K. I'll take a look at that today. Stay tuned.
Cheers, - Takashi
Hi Bert,
I published Korakurider's new change set and mine.
1464TransExport3-KR.cs 1465TransExportUI-tak.cs
As you said, pot / po files are divided per top category except Morphic. How do you feel? It's ok but looks like still too much for me...
To generate gettext template files,
open - Language Editor
To export template: gettext - gettext template.
To export translation (split files): gettext - gettext export
To import translation: gettext -gettext import.
Although, here are still un-exported keyword e.g. Object Catalog etc. I'll fix it tomorrow, so please wait a while.
Cheers, - Takashi
Bert Freudenberg wrote:
Takashi would know better, but I'd hope very soon, a few days only. It's a high-priority task.
Not bad, thanks :)
We should create a better file hierarchy, named "po". In think the class category should be the directory name and contain the translations:
po/EToys/EToys.pot po/EToys/de.po po/EToys/pt.po po/Connectors/Connectors.pot po/Connectors/de.po po/Connectors/pt.po
Also, as you mentioned, we still should reduce the number of categories. Like, all Connectors stuff should be one file only.
And I notice that some Comments are not too helpful - like, all comments in Protocols.pot are the same.
- Bert -
On Jul 23, 2007, at 10:41 , Takashi Yamamiya wrote:
Hi Bert,
I published Korakurider's new change set and mine.
1464TransExport3-KR.cs 1465TransExportUI-tak.cs
As you said, pot / po files are divided per top category except Morphic. How do you feel? It's ok but looks like still too much for me...
To generate gettext template files,
open - Language Editor
To export template: gettext - gettext template.
To export translation (split files): gettext - gettext export
To import translation: gettext -gettext import.
Although, here are still un-exported keyword e.g. Object Catalog etc. I'll fix it tomorrow, so please wait a while.
Cheers,
- Takashi
Bert Freudenberg wrote:
Takashi would know better, but I'd hope very soon, a few days only. It's a high-priority task.
Hi Bert,
We should create a better file hierarchy, named "po". In think the class category should be the directory name and contain the translations:
po/EToys/EToys.pot po/EToys/de.po po/EToys/pt.po po/Connectors/Connectors.pot po/Connectors/de.po po/Connectors/pt.po
It is easy though, how do we actually setup the translation environment? If we follow Fedora's way, is it fit to Fedora's convention? I have looked http://wiki.laptop.org/go/Activity_Translations shortly, but I couldn't find out the exactly process we should to.
Or, how about to use launchpad.net? https://translations.launchpad.net/ It has fancy web tool for gettext so that it seems to be good for non hacker people.
Also, as you mentioned, we still should reduce the number of categories. Like, all Connectors stuff should be one file only.
How many files did you expect? Yoshiki suggested me that three is enough. It is hard to decide before getting started actual translation process.
And I notice that some Comments are not too helpful - like, all comments in Protocols.pot are the same.
I know, current one is just a hack. The gettext exporter only works with literal receivers properly. We should fix, but it will need more a week.
Cheers, - Takashi
On Jul 23, 2007, at 22:39 , Takashi Yamamiya wrote:
Hi Bert,
We should create a better file hierarchy, named "po". In think the class category should be the directory name and contain the translations: po/EToys/EToys.pot po/EToys/de.po po/EToys/pt.po po/Connectors/Connectors.pot po/Connectors/de.po po/Connectors/pt.po
It is easy though, how do we actually setup the translation environment? If we follow Fedora's way, is it fit to Fedora's convention?
The directory layout is pretty much independent of what the translators will see.
I have looked http://wiki.laptop.org/go/Activity_Translations shortly, but I couldn't find out the exactly process we should to.
Well the "process" just is to open a bug about translation and hope for people jumping on.
Or, how about to use launchpad.net? https:// translations.launchpad.net/ It has fancy web tool for gettext so that it seems to be good for non hacker people.
Yes, I like that a lot, we should use it if we can, and announce on squeak-dev once we did so. OTOH, we would need to get the OLPC version of etoys into that system first, I'm not sure how simple that is - it's not just a translation site but a whole project hosting site.
Also, as you mentioned, we still should reduce the number of categories. Like, all Connectors stuff should be one file only.
How many files did you expect? Yoshiki suggested me that three is enough. It is hard to decide before getting started actual translation process.
Well ... for OLPC we only really need one file, true. However, I suspect a file with a couple of thousand entries might scare contributors - that's one reason to split. For "other" software you have a separate translation file for each library and each application. That's the granularity people expect, a few hundred entries max. Squeak combines so many things into one that a single translation file becomes unwieldy. Also, separate files give context. And ultimately we don't want to maintain these translations on our own but let the Squeak community handle it.
That's why I think we need one file for everything that could reasonably be "broken out". That matches quite well with class category, given that's the granularity that Monticello uses and people are accepting. I have attached the count of translations per file as of now for reference.
What we could do to not end up with too many files for now is have one "catch-all" fallback file and then a few files based on class category. I'm open for suggestions though :)
And I notice that some Comments are not too helpful - like, all comments in Protocols.pot are the same.
I know, current one is just a hack. The gettext exporter only works with literal receivers properly. We should fix, but it will need more a week.
Well, if it takes that time so be it. We need to get it right. Btw, I think the vocabulary translations belong to the classes they are declared in, rather than in a central file.
- Bert -
657 Protocols.pot 218 Morphic-Kernel.pot 181 Morphic-Scripting.pot 144 System.pot 137 Sound.pot 119 Morphic-Experimental.pot 101 Morphic-Worlds.pot 90 Morphic-Books.pot 80 Morphic-Widgets.pot 76 Morphic-Stacks.pot 71 Tools.pot 71 Morphic-Basic.pot 70 Morphic-Games.pot 68 Kernel.pot 54 Connectors.pot 52 Scamper.pot 52 Multilingual.pot 49 Movies.pot 47 Morphic-Navigators.pot 45 Morphic-Windows.pot 44 Morphic-Flaps.pot 41 Morphic-Mentoring.pot 40 ConnectorsShapes.pot 39 VideoForSqueak.pot 39 ConnectorsTools.pot 37 Morphic-Demo.pot 28 Balloon.pot 24 Morphic-Scripting Tiles.pot 23 Morphic-Support.pot 18 WS.pot 17 Sugar.pot 14 Morphic-PartsBin.pot 14 Morphic-Components.pot 13 Morphic-Palettes.pot 13 Morphic-Menus.pot 13 Graphics.pot 13 EToys.pot 11 StarSqueak.pot 8 Morphic-GeeMail.pot 8 ConnectorsText.pot 7 Nebraska.pot 7 Collections.pot 6 Network.pot 5 Morphic-Undo.pot 5 Morphic-Buttons.pot 4 ST80.pot 4 Proxy.pot 3 ProjectsRepositoryClient.pot 3 Morphic-Text Support.pot 2 Morphic-Scripting Support.pot 2 Morphic-Postscript Canvases.pot 2 Morphic-PDA.pot 2 BroomMorphs.pot
Just a quick report. I measured how much keywords are covered by new gettext exporter.
Detected keywords by GetTextExporter2: 2630 All known phrases in NaturalLanguageTranslator: 4526 Coverage: 58.1087052585064%
| categories keys diff | categories := Dictionary new. GetTextExporter2 new appendStringReceivers: categories. GetTextExporter2 new appendVocabularies: categories. keys := categories values inject: Set new into: [:set :next | set addAll: next keys; yourself]. diff := NaturalLanguageTranslator allKnownPhrases keys difference: keys. Transcript cr; show: 'Detected keywords by GetTextExporter2: ' , keys size printString. Transcript cr; show: 'All known phrases in NaturalLanguageTranslator: ' , NaturalLanguageTranslator allKnownPhrases size printString. Transcript cr; show: 'Coverage: ' , (keys size / NaturalLanguageTranslator allKnownPhrases size * 100.0) printString , '%'. diff inspect
Although, All known phrases includes much unused keywords.
Cheers, - Takashi
And I notice that some Comments are not too helpful - like, all comments in Protocols.pot are the same.
I know, current one is just a hack. The gettext exporter only works with literal receivers properly. We should fix, but it will need more a week.
Well, if it takes that time so be it. We need to get it right. Btw, I think the vocabulary translations belong to the classes they are declared in, rather than in a central file.
Or, how about to use launchpad.net? https:// translations.launchpad.net/ It has fancy web tool for gettext so that it seems to be good for non hacker people.
Yes, I like that a lot, we should use it if we can, and announce on squeak-dev once we did so. OTOH, we would need to get the OLPC version of etoys into that system first, I'm not sure how simple that is - it's not just a translation site but a whole project hosting site.
I'm going to study more about this site.
How many files did you expect? Yoshiki suggested me that three is enough.
Well ... for OLPC we only really need one file, true. However, I suspect a file with a couple of thousand entries might scare contributors - that's one reason to split.
Yes, I agree. Let's try along the line.
And I notice that some Comments are not too helpful - like, all comments in Protocols.pot are the same.
I know, current one is just a hack. The gettext exporter only works with literal receivers properly. We should fix, but it will need more a week.
Well, if it takes that time so be it. We need to get it right. Btw, I think the vocabulary translations belong to the classes they are declared in, rather than in a central file.
Korakurider was saying he had new idea.
And I'm trying a way like gettext-noop() for the object catalog. http://www.gnu.org/software/gettext/manual/html_node/Special-cases.html#Spec... It seems to work better than I thought.
Cheers, - Takashi
etoys-dev@lists.squeakfoundation.org
-
Bert Freudenberg -
Takashi Yamamiya -
Tibi