On Thu, 2013-06-20 at 22:49 -0400, Rahul Sundaram wrote: > On 06/20/2013 10:05 PM, Adam Williamson wrote: > > So if I'm reading things right, no, langtable doesn't currently make > > any attempt to associate each Indian language with a territory any > > more specific than 'India', and as currently implemented, couldn't > > actually do this. Mike may well correct me if I'm wrong and he's > > reading, though. It may be the case that he'd consider it valid to > > extend the concept of a 'territory' down to province/state level but > > just hasn't implemented it yet, or it may be that he considers it to > > be strictly tied to the ISO 3166 standard. I don't know. > > https://en.wikipedia.org/wiki/Languages_with_official_status_in_India#States > has a mapping of languages to states if anyone wants to go that route > and if you need more help, feel free to ping me. So I did one more bit of digging into this - went and checked what anaconda actually *does* with prelocation and translations. Here it is, from welcome.py: # We can use the territory from geolocation here # to preselect the translation, when it's available. territory = geoloc.get_territory_code() self.language = Language(LOCALE_PREFERENCES, territory=territory) # check if there is one and only one locale for the territory if len(self.language.preferred_locales) != 1: log.info("Didn't get a single locale from Geolocation," " falling back to default locale.") self.language = Language(LOCALE_PREFERENCES, territory=None) # Explanation: # Some territories have multiple locales, # for example, the Switzerland has: # de_CH, it_CH and fr_CH # As there is no clear order of preference for them, # it is safer to just fall back to the default locale So it definitely is trying to pick a default language from the geolocation-determined territory, but if the territory it decides you're in has more than one language associated with it in langtable, it gives up and falls back on 'the default locale' (i.e. US English). So this is actually good for our big test case, India - presumably people in India will wind up with U.S. English as the pre-selected language, which is what we thought would be best anyway. There could potentially still be some problems, but at least some of the more problematic cases should actually be okay. -- Adam Williamson Fedora QA Community Monkey IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora http://www.happyassassin.net -- test mailing list test@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test