Hi all,
I have some more observations and ideas on the localization of the general GUI and possibly also related to multi lingual metadata.
I was merging new localized strings today and got !@!&^%$@#^&*%$ once more... 
Main reason:
The way we maintain strings is not very effective when looking at it from a maintenance point of view.
Related reason:
Everybody that adds new strings/ localizes new part, seems to take the easy route and dump the new strings at the start or end of a file in fairly random order.
Consequence:
Someone (and for now that has been me, hence my complaint!) has to go through all language files and manually sort them, remove duplicates, discover missing strings and possibly declare old ones as obsolete. I did this today for just one of them (Arabic, not used and so probably not the most useful one, but hey it is the first language in the list....)
This has become a sheer impossible task by now. Only the strings.xml file contains over 500 lines per language. Not to even mention the other 27 XML files in the loc folder (again per language), the loc XML files in each schema and the loc files in the InterMap application.
I started flaming on some IRC box, and had a discussion following on that that pointed me in the direction of a more standardized approach. There may be others so don't hesitate to mention those in response to this email.
The tools we could use are:
- XLIFF as internal format to store the language files in. These could than be exported into the format we actually need (but maybe those could be just the same!?).
- Pootl as the web interface to allow people to translate strings. We could start a discussion at OSGeo level to see if there are more projects interested to have this as a facility for more OSGeo projects.
I updated the localization discussion page I put up some time ago at i18n – GeoNetwork opensource Developer website and I hope you can have a look there and give feedback/ continue to discuss with me what solution we should go for in the future.
We can, with experiences from the encoding used in XLIFF (or other i18n practices) also discuss how to best tackle the multi lingual metadata content. Maybe some Pootl concepts could be used, but I'm saying this while completely ignorant on those concepts
One thing is sure, I need some people that will take the same tedious task as I did to go through the strings.xml files and put elements in alphabetic order, remove dup.... etc....
Ciao,
Jeroen
