Evan Prodromou wrote:
[about merging the Wikipedia DBs into one DB]
I think there are a number of issues:
- Marking articles with a language code. That's pretty straightforward, I think: add another column for the language code.
My suggestion was to use this opportunity to migrate to an entirely new database schema, which in my view will allow us better performance in the future: http://meta.wikipedia.org/wiki/Experimental_new_database_schema
- User interface language. I believe this can be determined from the browser headers, and we could have an override set in a cookie or a user preference.
My idea here was to display the UI in the same language as the article if you are anonymous, and to use a user preference if you are logged in.
- Choosing which version of an article to show. This is easy if the there's a version in the user's language
The subdomain determines this (e.g. fr.wikipedia.org as opposed to es.wikipedia.org, say).
- Cultures. This is the really hard one. It'd be difficult to integrate the N Wikipedias into one. It's hard to do decision-making in multiple languages.
That's entirely separate from merging the DBs. I'm not sure if merging the "cultures", as you call them, is even possible, but that doesn't stop the DBs from merging.
Timwi