Hi,
I'm maintainer of specspo package. I regenerated specspo pot files and slightly changed tree structure.
The process looked like this:
- generate new po catalgue from selected rpm repos (usually
rawhide-nightly)
- commit the changes on ELVIS cvs tree
- translation team hard-work here
- after they commit their changes put everything to tarball and update
specspo on DIST with it
- wait till next release (FC or RHEL) and goto 1
This shows we had currently just one specspo ELVIS upstream for
everything we release. This specspo included everything I decide to
include according to what we are about to release.
Problems:
- it was not possible to regenerate catalogues for recent release. If
package descr or summary changed for recent product we were not able to
do the translation (or at least I didn't know how)
- upstream on ELVIS was not updated, it was recreated from scratch (from
different sources) - this can make possibly troubles to translation team (? - correct me if I am wrong)
- we had one specspo for everything (this could contain a lot of packages
not included in dist).
To solve some of these problems I decided to split current specspo tree into branches like DIST tree is.
So since now it looks like this:
|-- common
|-- RHEL-5
| `-- dist
| |-- desc
| `-- summary
`-- devel
`-- dist
|-- desc
`-- summary
Directory common contains some service scripts. Under RHEL-5 and devel is same structure you are used to.
After release of fc7 I make copy of devel called FC-7.
I am about to improve service scripts so they can post mail to fedora-trans-list if pot files in directory other then devel are changed. (this should by VERY rare).
I hope that than the process remain unchanged for translation team (you just need to keep translation in specspo/devel/ up to date).
If I can do anything that can make your translation of specspo smoother, let me know.
Also any comment or idea would be appreciated.
Regards,
Martin