Some people get confused by those partially translated *.properties file in Gaia tree and think they are official l10n support for those languages. In fact, these files are just for test purpose to check localized string can be displayed correctly. But partially translated locales sometimes won't help on this test but fool a few developers to think a string isn't been localized. So I file this bug to discuss a better mechanism of testing l10n stuff, like: using generated upside down English(en-UD, proposed by Tim). But I know we choose these three languages with other special meanings: 1. test right to left text (ar.properties) 2. test longer text (fr.properties) 3. test Square-Block Characters (zh-TW.properties) So maybe en-UD is not the only one approach here.
I want to hear what Stas feel about the proposal before we continue, as he was the original person who check-in these locale files. Also, I didn't invent the en-UD locale "English (Upside down)" -- Facebook did!
I have a work-in-progress patch for pseudo localization in bug 900182. The work there has the benefit that it's actually done at build time, so developers can just use that for their in-progress patches. The idea is to create various lengths as well as a RTL pseudo locale. I don't have a good algorithm for a pseudo-zh-*, though. Not sure how much value we actually got out of this one. I'm not yet sure what to do with the in-tree localizations once we have working pseudo l10n in the tree, we may just want to remove them? Not sure if this bug is a dupe?
That's probably not a dup but a different approach to solve the same issue. IMO we could -- in this bug, remove the outdated locale. -- in bug 900182, create fake locales.
WONTFIX since bug 900182 already shipped with fake locales, and source repo l10n file to be removed in bug 1011519.
Status: NEW → RESOLVED
Last Resolved: 4 years ago
Resolution: --- → WONTFIX
Mass Edit: adding the [rtl-meta]
QA Whiteboard: [rtl-impact]
You need to log in before you can comment on or make changes to this bug.