The SQL language is terrible but we can live with it.
But the answer to "Are there any relational data stores that offer eventual consistency, easy distribution, schema-on-demand or any such things a large modern application can use?" appears to be no. And that's just awful.
Dear Important People: would the PostgreSQL project consider supporting other query languages? Or creating a plug-in mechanism for them, so that alternative interface languages could be added without changing the base code?
If by important you mean possessing a commit-bit then I don't count...but for me, such a project would have to gain significant adoption as a fork of the PostgreSQL code base before it would ever be considered for take-over by the mainline project.
