> I'm most familiar with relational databases and the way in which their data is stored is generally dictated by the schema.
One of the major features of relational databases (and the improvement over their predecessors) was precisely the disconnect between logical and physical model. If the logical schema dictates how something is stored physically, then it's not a terribly good system. At the very least the intention with relational databases was to provide for this extra flexibility that previous models didn't have. Sure you don't have to exercise that flexibility but it's always been there.
> we'd be talking about a database that accepts arbitrary groupings of predefined sets of fields, which then on-the-fly determines how best to pack these individual sets together with other sets belonging to other groupings
So basically columnar storage with table inheritance would be enough to do the trick. I know there's databases for either of those features, and it's absolutely not a huge leap to put the two together. Fact is I wanted to implement something like this for CLOS but didn't have the time yet, sadly.
One of the major features of relational databases (and the improvement over their predecessors) was precisely the disconnect between logical and physical model. If the logical schema dictates how something is stored physically, then it's not a terribly good system. At the very least the intention with relational databases was to provide for this extra flexibility that previous models didn't have. Sure you don't have to exercise that flexibility but it's always been there.
> we'd be talking about a database that accepts arbitrary groupings of predefined sets of fields, which then on-the-fly determines how best to pack these individual sets together with other sets belonging to other groupings
So basically columnar storage with table inheritance would be enough to do the trick. I know there's databases for either of those features, and it's absolutely not a huge leap to put the two together. Fact is I wanted to implement something like this for CLOS but didn't have the time yet, sadly.