Berto 'd Sera wrote:
Hoi!
. This could be "guessed at" by looking at existing, incomplete files. Probably there is a fairly predictable order to which messages get translated first.
You're a genius :) Yeah, we can probably get a pattern out of that.
So..
If we have a macro to load them up in a MySql db we get a 3-vectored space:
- lancode (PK1)
- message name (PK2; this is supposed to hold unchanged thru localization)
- text
We can get very close to "reading people's vote" by seeing which msgs are localized say by at least 75% of the existing interfaces.
Yeah... al it should take is a php script that would parse a directory of message files and load them in the table. After that we simply run some "SELECT COUNT(*)"s to generate all the comparative analysis we need.
If the $msg_names hold unchanged it shouldn't even be a long script, because you can simply parse the existing variables and use their name/content pair without writing any CASE statement at all, right?
I have made a little list as r20920 of the messages at http://www.mediawiki.org/wiki/User:Platonides/TranslatedVars
Not surplrisingly, the most translated are thinks like mainpage, help or mytalk, needed to have a good interface.