* Betr.: " Re: [tryton-dev] About some issues faced during the last release process" (Tue, 8 May 2012 00:45:55 +0200):
> On 08/05/12 00:19 +0200, Mathias Behrle wrote:
> > * Betr.: " [tryton-dev] About some issues faced during the last release
> > process" (Mon, 7 May 2012 21:08:45 +0200):
> >
> > > I would like to report some issues I faced during the last release:
> > >
> > > - format of the po file
> > > Some have pushed files not generated by the tryton export which means
> > > it was not wrapped correctly.
> > > This is wrong, we must generate files always the same way to minimize
> > > the changeset diff. This is important when you search in the history.
> > > (exception of the client which uses Babel).
> >
> > Of course, I am prefering minimzed changesets as well.
> > OTOH I see, that some contributors are assigning copyright to their
> > translations - something not provided so far by trytond itself. What's your
> > idea of a common export interface?
>
> I don't know. I even don't know if it could be considered as an original
> work.
> I would like to have the same behavior on the client side than on the
> server but it is an hard work.
>
> > > - multiple commit on the same file
> > > During the translation window, translators must commit only once the
> > > translation of a module. We don't need to have WIP in the repository.
> > > The repository is not a tool to send WIP files. Again we must have a
> > > clean history and try to reduce the number of changeset on a file.
> >
> > I don't agree. All the code is work in progress. I for my share try to push
> > translations as early as I am able to do, thus enabling possible testers to
> > also check the translation. And of course I am trying to minimize fixes
> > after the first push (in fact for 2.4 I only had to update one translation
> > due to a fix pushed by upstream during string freeze), but a fix is a
> > fix ;), and it is the privilege of translators to push their fixes during
> > string feeze.
>
> No, it is not the way we work on Tryton, remeber codereview, bug
> tracker, mailing list etc.
> Pushing stuff for testing is *wrong* and it is the best way to break
> stuffs.
What's the difference between pushing code or translation? It is never meant to
be changed, but if it has to be fixed...
> If you want to share, share it somewhere else.
Or lose contributors...
> > > - translation outside the release window
> > > This is just WIP commits because the strings are not freezed so it is
> > > not needed to run after them.
> > > If some translators find they don't have enough time (3-4 weeks)
> > > during the release window, they can start (or maintain) outside the
> > > main repository an up to date translation as far as it doesn't trouble
> > > the history of the main repository.
> > > Also we can try to setup a transifex instance on tryton.org if it is
> > > needed (need to be patched to remove GNU headers).
> >
> > Translation is development, the same way as code is. There has to be an
> > *easy* way to test translations. So far the only way is to push to the
> > repos.
>
> See above.
See above.
--
Mathias Behrle
MBSolutions
Gilgenmatten 10 A
D-79114 Freiburg
Tel: +49(761)471023
Fax: +49(761)4770816
http://m9s.biz
UStIdNr: DE 142009020
PGP/GnuPG key availabable from any keyserver, ID: 0x8405BBF6
signature.asc
Description: PGP signature
