A better way of doing templates?
Jordi Mallach brings up a good point on the rosetta-users mailing list, in response to my queries there.
Rosetta doesn't really work with non-po files, so the fact that we have templates that need translating will cause us some grief. Well, even more grief than it has over the years. ;)
Does anybody have any good ideas for handling this in the Mailman 2 tree? For MM3, it's a top priority that any templating system we use must support po-ification for i18n. For Mailman 2.2 it's more difficult because I don't think we should adopt an entirely new template system. It's even worse for 2.1 because we're not accepting new features, although if we came up with something simple and minimally disruptive, I'd consider it for the benefit of improving the translation process.
Ideas are welcome! -Barry
Hi,
Barry Warsaw wrote:
Jordi Mallach brings up a good point on the rosetta-users mailing list, in response to my queries there.
Rosetta doesn't really work with non-po files, so the fact that we have templates that need translating will cause us some grief. Well, even more grief than it has over the years. ;)
Does anybody have any good ideas for handling this in the Mailman 2 tree?
How about adding current templates into pot/po and pickup and generate language templates during "make install." I don't know if it could be done at all but I will look at it this weekend.
As regards the whole translation process, I've already submitted a patch to select languages when configuring. This should help separating the translations from the main developement code base. http://sourceforge.net/tracker/index.php?func=detail&aid=1298355&group_id=103&atid=300103
-- Tokio Kikuchi, tkikuchi@ is.kochi-u.ac.jp http://weather.is.kochi-u.ac.jp/
On Thu, 2005-11-24 at 10:44 +0900, Tokio Kikuchi wrote:
How about adding current templates into pot/po and pickup and generate language templates during "make install." I don't know if it could be done at all but I will look at it this weekend.
I see that this is checked in, which I guess means that the idea is working well for you. Cool! If we're going to generate all the template files at build time, I suppose we should remove all those files (and the directories?) from CVS. It's generally not a good idea to check in files that are generated during the build process. What do you think?
As regards the whole translation process, I've already submitted a patch to select languages when configuring. This should help separating the translations from the main developement code base. http://sourceforge.net/tracker/index.php?func=detail&aid=1298355&group_id=103&atid=300103
Very cool. I've also heard requests to reduce the tarball size by separating the languages out into separate downloads. Does anybody think that's worth doing?
-Barry
Barry Warsaw wrote:
On Thu, 2005-11-24 at 10:44 +0900, Tokio Kikuchi wrote:
How about adding current templates into pot/po and pickup and generate language templates during "make install." I don't know if it could be done at all but I will look at it this weekend.
I see that this is checked in, which I guess means that the idea is working well for you. Cool!
Yeah! ;^)
If we're going to generate all the template files at build time, I suppose we should remove all those files (and the directories?) from CVS. It's generally not a good idea to check in files that are generated during the build process. What do you think?
I'm going to do this as a next step soon.
As regards the whole translation process, I've already submitted a patch to select languages when configuring. This should help separating the translations from the main developement code base. http://sourceforge.net/tracker/index.php?func=detail&aid=1298355&group_id=103&atid=300103
Very cool. I've also heard requests to reduce the tarball size by separating the languages out into separate downloads. Does anybody think that's worth doing?
This has also been checked in. I'm going to try hack the release script to separate the languages from the main tarball.
Cheers,
-- Tokio Kikuchi, tkikuchi@ is.kochi-u.ac.jp http://weather.is.kochi-u.ac.jp/
Very cool. I've also heard requests to reduce the tarball size by separating the languages out into separate downloads. Does anybody think that's worth doing?
This has also been checked in. I'm going to try hack the release script to separate the languages from the main tarball.
I personnally dissent from this idea; one great strength of free software is its internationalization open to everyone both in participating to translations and using them. Making them "optional" is making the software weaker.
If the need for smaller tarballs is real (which I doubt, given the bandwidth and diskspace you need to run Mailman :-) ), one option could be, on the contrary, to have a seperate download link for people who *want* a mono-lingual (supposedly US-English?) Mailman.
-- Fil
On Wed, 07 Dec 2005 09:20:06 +0100, Fil <fil@rezo.net> wrote:
Very cool. I've also heard requests to reduce the tarball size by separating the languages out into separate downloads. Does anybody think that's worth doing?
This has also been checked in. I'm going to try hack the release script to separate the languages from the main tarball.
I personnally dissent from this idea; one great strength of free
software is its internationalization open to everyone both in participating to translations and using them. Making them "optional" is making the
software weaker.If the need for smaller tarballs is real (which I doubt, given the
bandwidth and diskspace you need to run Mailman :-) ), one option could be, on the contrary, to have a seperate download link for people who *want* a mono-lingual (supposedly US-English?) Mailman.
May I suggest a zero-lingual base version? The only way to make software
truly language independent is to avoid the concept of a "default language".
If everybody has to get a language file, developers will be forced to
think about localization issues.
-- Using M2, Opera's revolutionary e-mail client: http://www.opera.com/m2/
At 1:08 PM +0100 2005-12-07, =?utf-8?Q?Bj=C3=B8rn_Vermo?= wrote:
May I suggest a zero-lingual base version? The only way to make software truly language independent is to avoid the concept of a "default language". If everybody has to get a language file, developers will be forced to think about localization issues.
You have to use some sort of tag to translate the message into, and that tag is going to have to exist in some language. And the programmers are going to have to think in some language while writing the code which will handle the tag.
Or do you propose that all the programmers are going to have to forget all the native languages they know and instead use randomly generated gibberish tags? How do you propose to maintain that code?
When you go talking to other people in the world, do you work in a zero-language manner internally, so that you can make sure that you never have to translate from some native language into another native language?
When you can do that for all your native language discussions, you can come back here and share with us how that works and how you would modify that to work with programming languages and i18n.
-- Brad Knowles, <brad@stop.mail-abuse.org>
"Those who would give up essential Liberty, to purchase a little temporary Safety, deserve neither Liberty nor Safety."
-- Benjamin Franklin (1706-1790), reply of the Pennsylvania
Assembly to the Governor, November 11, 1755LOPSA member since December 2005. See <http://www.lopsa.org/>.
On Wed, 2005-12-07 at 13:08 +0100, Bjørn Vermo wrote:
May I suggest a zero-lingual base version? The only way to make software
truly language independent is to avoid the concept of a "default language". If everybody has to get a language file, developers will be forced to
think about localization issues.
That's difficult. IIUC what you propose is a more 'catgets' style than a 'gettext' style of coding. E.g. where each human readable message in the source is given a number so that even English messages must be looked up in the catalog. The alternative is what Mailman currently uses -- the English message serves as the lookup key in the catalog.
Ages ago we experimented with both styles. I highly dislike message numbers and prefer gettext, although admittedly I'm probably biased by being a native English speaker.
As for Fil's opinion on splitting the distribution up by languages, I kind of agree. I think we should continue to produce fat distributions with all languages, possibly providing an English-only (i.e. no catalogs) distribution as a thin alternative. I'm still +1 though on Tokio's changes to allow the site admin to /install/ Mailman with fewer languages.
-Barry
participants (5)
-
Barry Warsaw -
Bjørn Vermo -
Brad Knowles -
Fil -
Tokio Kikuchi