After playing a bit with Pootle, here are my conclusions:
- Pootle can work with java .properties files, but this is tricky. We
need to convert between .properties and .po files;
- it is possible to make pootle work with subversion;
- possibility to perform automatic consistency checks on translated
files, to ensure quality of translations. This is for example the
capitalization of words or the length of string. The complete list of
checks is available at ;
- we have the possibility to let people suggest better translations;
- provide possibility to easily form translation teams and set goals;
- however, the interface for current version of Pootle is not intuitive.
This means we need to document extensively how to participate in the
Our project can really benefit from an online translation platform, but
the use of Pootle for SIP Communicator is not straightforward because of
java .properties files. Here is a possible solution.
The pootle server need to have a running job that fetch new translation
template (our resource.properties) regularly from svn and that merge its
changes (new key or modified text) into converted .po language files.
Any change in the resource.properties file on the svn server will be
propagated and visible in all translation on the pootle server via the
web interface. At this point, translators can work on new strings via
the Pootle web interface.
Translators can commit their changes through pootle. However, .po files
first need to be converted to .properties files before they can be used
in SIP Communicator. We have the choice between:
1. convert .po files to .properties before the commit by Pootle. This is
currently not automatic, but upcoming Pootle release seams to handle
that the case.
2. store translated files as .po in our svn, and convert at build time.
On a final note, the upcoming release of Pootle seems to address a lot
of usability problems. It also bring the ability to convert files before
they are commited to a source repository. But, they completely their
architecture (now using a database) and I have no idea how difficult it
is to migrate from current (1.2) to upcoming version (1.3).
We could either try to go with Pootle 1.2 knowing that migration to 1.3
may be time consuming, in this case we need to decide how to store
translation files on our svn. Or we could wait until Pootle 1.3 is out
(no clue about a possible release date).
What do you think?