Alexandre Prokoudine scrisse:
Here is an question that is triggered by a recent request in inkscape-translator@...233...
IIRC, this was triggered for a different reason (too large file difficult to handle on old hardware). In that case I think we agreed that we don't really need it. This just to point out the source. What you say here is however quite different; see below for a discussion of your proposal.
While doing refactoring in 0.47, should we probably try to split translatable messages into several ones? E.g.: Inkscape has grown *a lot* over the 0.46: amount of translatable messages has jumped from 2480 in 0.45.1 to 3465 in 0.46. While supporting one large already existing PO file is probably OK for translators (not quite correct for the case I started this email with), creating a whole new translation will be psychologically difficult at the same time. I've seen this with Xara Xtreme for Linux which had ~5700 messages last time I checked - and the pace of translation work into various languages was dead slow.
I think you're missing two different things. Regarding these lots of string, it is true, but I've already seen (and worked on) project with similar/bigger number and it is quite normal for "nice" project. Plus, we historically accept po with a near-zero translation ratio (like the old sodipodi legacy) we commit them quickly, and we'd encourage l10n in all way (news item on main site, allowing partial translation, a friendly and populated ML and so on). I really don't know the community behind Xara, but if you see some differences between the two l10n management, these as well reflect on translation work.
I really prefer to keep our strings in a single .po, as it will help coherence, keep the mechanism simple, avoid useless duplication of work and won't burden i18n management.
Anyway, I think in the end we'll have to split some strings. I'm speaking of extensions, which will probably be decentralized (and I'm not yet sure how we'll handle their translations). Plus the day we'll come with a separate set of libraries (and maybe pedro), then the split would naturally happens.
For now, I think we should stay "as is".
Alexandre
Ciao, Luca