[Fle3-dev] Migrating translation files to .po format

[Tarmo Toikkanen]

Our current translation system sucks. Translators need to use a
special editor and make sure that encoding, line breaks and quotations
aren't broken. I propose that we migrate to .po files.

.po is the de-facto standard of translating strings in the open source
community. There are several translation programs (like gtranslator
and kbabel) that take care of all the syntactic hassles. The
translator (human) needs only to do the translations. Also, any
changes or updates to the strings can be marked as "fuzzy" to draw the
translator's attention.

The problem is that .po files are designed for C, where original
strings are processed at compile time. This won't work for our
Zope/Python product. Holding all 500 translations in a hash table
indexed with the original long english strings is too much of a
burden.

So the requirements are:

- In the runtime system we hold only short keywords (like now) and
  their translations to all languages.
- The translation files are in .po format (where each original string
  is paired with a translated string).
  
We will need a utility to generate .po files from the original
vocabulary.py file. Translated .po files can either (A) be converted to
.py format or (B) read into memory when the system reloads the
translations.

So... Any comments or ideas, before I start implementation? Or does
anyone else want to implement this? I do have other things I can do,
you know... ;)

--
Tarmo Toikkanen
The Fle3 Team
http://fle3.uiah.fi/