The localization process will be changed into a continuous process soon

I am happy to announce that soon the localization process will be 
changed into a continuous process which will allow to provide l10n 
turnarounds cycles at every milestone. This process was presented at the 
OooConf In Budapest: 
http://www.ooocon.org/index.php/ooocon/2010/paper/view/207 and is the 
outcome of Gregor, Rafaella and myself working on the so called 
"continuous l10n" project.

Here below the major steps of this workflow:
*it creates a cws named l10n<workstamp><milestone> , e.g. l10ndev300m94
*it downloads all l10n from pootle
*it downloads  all sdf found on 
ftp://qa-upload.services.openoffice.org/l10n/upload/ (no more sdf 
delivery via Issue. Translated sdf files can be uploaded to this ftp)
*it gsichecks and merges the translated sdf files. Precondition is that 
language is known and is set up in the resource system.
*it builds languagepacks for all languages that have been changed (even 
when only 1 string has been changed) + one en-US insetsets
*it removes automatically languages from the build which cause a build 
error not covered by the gsicheck, this allowing the other languages to 
be built.
*it uploads the *_err.sdf from gsicheck
*it uploads windows and linux32 languagepacks and the en-US insetsets to 
ftp://qa-upload.services.openoffice.org/l10n/download/

Please help us testing this new process! How?
*Check the language packs created by this process
*Non-Pootle users should upload translated sdf files to 
ftp://qa-upload.services.openoffice.org/l10n/upload

For a later stage we are planning to
* commit the l10n
* integrate every cws
We are working already on a technical solution that will not blow up the 
main repository. We will not integrate those cws till we solve this issue.

Cheers,
Ivo