OK, now I see why it is called 'dumb'; the thing literally just dumps
out the repr of two strings on each line for key/value pairs. To read it
just evals each line in the string. And whichdb detects this format by
looking for ' or " as the first character since that is what the repr
for str is. And that is why the Latin-1 decoding was used; it has the
proper repr for write-out.