El Dimecres, 19 de novembre de 2014, a les 01:30:23, Bruno Coudoin va=20=

escriure:
> Le 18/11/2014 12:43, Burkhard L=FCck a =E9crit :
> > What about http://gcompris.net/wiki/Translation_addons and
> > http://gcompris.net/wiki/Word_Lists ?
> >=20
> > The Legacy Gtk+ Version had some resources in xml files to translat=
e, e.
> > g.
> > wordsgame.xml, gletters.xml, readingv.xml and hangman.xml
> >=20
> > In the new gcompris this seems to be done via json files, e. g.
> > activities/gletters/resource/default-[lang].json
> >=20
> > Do we need to integrate the translations of these resources into sc=
ripty?
>=20
> Hi,
>=20
> Yes, I have not yet mentioned about that but this is the right time t=
o
> talk about it.
>=20
> Beside the regular po file we have dataset files for several activiti=
es
> that have are localized instead of translated. As you mentionned it,
> these files have been translated from xml to json file during the por=
t.

How do you see the translation of those datasets happening? I.e. do you=
 want=20

a) The translation is in the json
b) The translation is in the .po file

b) is certainly easier, since you just need to extract the text from th=
e .json=20
to a dummy cpp in Messages.sh time and then just call=20
i18n(theTextInEnglishFromTheJson) and everything works.

Cheers,
  Albert

>=20
> I don't know what is the best i18n strategy to adopt, keep them in
> GCompris or moving them into scripty. I it helps in the decision, as =
an
> input, these files once done are usually pretty stable over time.
>=20
> As it will require some work on my side and on yours to have then und=
er
> scripty, I would like to go this way only if there is a real added va=
lue.
>=20
> We have not yet mentioned it but this is an important part in GCompri=
s,
> we also have a large dataset of voices. This represent about 200MB in=
 40
> languages. In the Gtk+ version it was in the same git as the code but=

> this is a huge mistake. It has to be put is a separate repo and svn
> works better for this kind of workload. Do you have some suggestion o=
n
> the best place for this?
>=20
> Bruno.