Re: [libvirt] l10n/i18n ~ po files and their cycle

On 26/01/2011, at 7:58 AM, Zdenek Styblik wrote: <snip> I *think* the rough idea with the translations for things, is that there is a web interface for doing translation stuff. The software used is called "Transifex": http://www.transifex.net/ The way it kind of works, is that when you have a new project (call it "myfoo" for example), you can make sure all of the "master" text strings needing translation are put into one file. (call it "myfoo.pot") You then register the myfoo project with the Transifex crowd, pointing them to the source code repository, and the myfoo.pot file in it. Transifex reads the text strings from that file. (git://www.myfoo.org/source/myfoo.pot) Transifex then have a web interface, where people can choose their language (say French) and it gives them a list of each text string from that master list. When a person then submits the text string, Transifex saves it to its internal database. At some point (generally before a new myfoo release) someone from the myfoo project logs into Transifex and gets a tarball dump of all the translated strings. These are in a format called po, and it's one per language. i.e. myfoo_cz.po, myfoo_es.po, myfoo_fr.po, etc The person adds these strings to the myfoo source code, and they get included with the release. Iin theory, from this point GNU gettext makes sure that when a user of myfoo has a locale with (say) French, then the French translations are loaded and displayed instead of those from the master language. Note, this is all a really rough blow by blow concept description. There are alternative software packages out there to Transifex, and there are also alternative translation file formats other than .pot/po/etc. But, it gives the idea. Does that help at all? :) + Justin