On Nov 23, 2016 00:47, "MF-Warburg" <mfwarburg@googlemail.com> wrote:
>
>
>
> 2016-11-22 15:33 GMT+01:00 Milos Rancic <millosh@gmail.com>:
>>
>> I don't think it's true at the moment, but imagine the next integration:
>>
>> * A person is born on year/January/date. That's the data Wikipedia
>> should take from Wikidata.
>> * A user says "I am a German from Germany" and has that as
>> localization, instead of default Austrian version.
>> * What's the method of telling Wikidata to give German German January
>> instead of Austrian German January inside of the infobox?
>
>
> Well, as dates in Wikidata are not stored as "5. Jänner 1980" in the first place, that seems no problem. The infobox' code will simply translate 1980-01-05 differently, depending on the users' language settings. Or am I mistaken?

That was just an example, not the best one. The point is that Wikidata operates with the open set of words and that we could easily come into the position to force a user to read even something completely strangr to him or her.

For example, the term Art Noveau/Secession and similar could easily become a category and a difference between the two varieties. And by reading one variety, a user could come into position not to understand that.

I could find a lot of such potential pairs between Serbian and Croatian, which are distant on similar level as Spanish varieties, so it's not hard to me to imagine that keeping strict ISO 639-3 codes instead of BCP 47 could make confusion.