Sam Johnston <samj at samj.net> writes: > I'm leaning towards the time/space tradeoff of including the ID in each > row somehow (in which case parsing into a hash of hashes is trivial > again). That works. Cheers, Chris.