You could be right, I only went skin deep.
> The way I read this table, what it says is that tokyo-cabinet (the database)
> is almost as fast as python-dicts, in memory...
>
> That seems to make an arguemnt to move to (some?) column versoin of this
> db... when looking also at things like Cassandra and Voldemort, this (to
> me at least) is making an additional argument to make a DAL for column-based
> databases - perhaps these 4 (Tokyo Cab, Cassandra, Voldemort, bigtable).
Interesting idea. Do any other frameworks support column-based DBs?.