[Maposmatic-dev] Need UTF8 encoded language files to proceed with Translation management at Transifex

[Guttorm Flatabø]

http://www.openstreetmap.org/user/Guttorm%20Flatab%C3%B8/diary/16804Hi all,

it's been looking like you're doing fine without Transifex for translations, and I've been busy with adding Waymarked Trails to Translatewiki, so I haven't done anything with MapOSMatic on Transifex until today.

However, trying out the Transifex workflow I bumped into some things that I'm hoping you can address:

1. The gettext template file has UTF8 characters outside ASCII encoded as escaped octals, f.ex. "\302\251", while the German translation uses "true" UTF8 encoding, f.ex. "©". Not only are the octals confusing to translators, they also confuse Transifex since the msgid in the template and translation aren't the same. From what I've gathered before, the octals aren't necessary and could be replaced with proper UTF8 characters. If that is not the case, then at least the template and translations need to use only either one of them so that msgid's match.
2. AFAIK there are no country varieties of Asturian, it is thus probably best to rename "ast_ES" to "ast", as ast_ES is a more specific code than necessary. This can be a real problem with language detection. F. ex. if you set Portuguese on the top of your language list in Chromium you will not be given the Portuguese version when you visit maposmatic.org because Chromium asks for "pt" and we can only offer "pt_BR". For Brazilian Portuguese this might be the intention, but for languages where there are no real country variety it just makes the chance of things going wrong higher. (It is not a problem the other way around. Try setting your Chromium language to "German - Austria", although maposmatic does not have an "de_AT" language, you'll still be given the "de" translation").
   As long as there are no "pt_PT" and "nl_BE" translations, I also think "pt_BR" and "nl_NL" should be renamed to "pt" and "nl" respectively and "tr_TR" should definitely be renamed to "tr".

Also, if you decide to use Transifex, you should beware of the following:

• The language file (.po) headers will be replaced, mostly for the better, but f.ex. "LAST-TRANSLATOR" content will be based on the translators who have contributed on Transifex, translators before Transifex will not be credited.
• Fuzzy translations and translation memory will be handled by Transifex and will thus be removed from the translation files.
• Maintenance is easy:
• Install the transifex client. Settings are stored in .tx
• The template should be automatically updated from git, but can also be pushed with: tx push -s
• Push new translations (should be an exception): tx push
• Get new translations from transifex: tx pull
• You should try to keep it simple and only do translations on Transifex.

So, hope we can get moving on this, because even though you seem to be doing well without I still think you'll be doing better with Transifex. I've commited most of these changes to my own fork at Github to illustrate. You can see what the translation files will look at in the last commit, but beware that it is corrupted by the UTF8 encoding problem.

--

Guttorm