[spambayes-dev] improving dumbdbm's survival chances...

[Skip Montanaro]

    Van> Hey, I'm nowhere near "Tim's sister" capability, but I still want
    Van> to just download the zip, extract it, and run the proxy. If dumbdbm
    Van> is a dumb way to go, it shouldn't be the default.

It's not the default.  It's currently the database of last resort.  If you
get dumbdbm it's because you don't have a working PyBSDDB (aka bsddb3) or
gdbm module and supporting library available.  Many Unix-flavored machines
have one or the other available out of the box.  Alas, the current
heavyweight champion of the world (Windows) doesn't come with anything
useful by default, so many people installing the Outlook plugin have been
getting the dumbdbm database module.

Mark, can you tweak the installer to try importing bsddb (and check for
bsddb.version) and/or gdbm, then emit a warning and fall back to using a
pickle if neither is available (or just tell the user where to get Berkeley
DB 4.x so the bsddb package can be loaded)?

    Van> I wouldn't be too upset to be retraining, I've only been running
    Van> this install for a week and could just start from scratch again. I
    Van> was planning on keeping up the training for a week or so anyway,
    Van> although my database is already up to 27 megs.

It would be a lot smaller if you were using PyBSDDB.  (That's another
dreadful bit about dumbdbm - each value stored consumes a minimum of 512
bytes.)  My database is only 21 megs and I've trained on over 20,000
messages.

Skip