BSDDB 1.8.x is buggy?

[Skip Montanaro]

    Warren> I have been exercising the bsddb unit with the following code.
    Warren> Can anyone show me something that will break bsddb? Or are the
    Warren> bugs too stealthy to produce a test case?

They are pretty stealthy.  There are bugs in the hash mode when you get
fairly large files.  I seriously doubt you'll notice them with files
containing just a handful of keys.

    Warren> The following test code starts with around 100 rows, and works
    Warren> it's way up to file sizes over 100 mb. If there were any
    Warren> duplicate key problems or core dumps in the database, would I
    Warren> find them this way?

Maybe.

    Warren> I have to ask myself, if code can work millions of times in a
    Warren> row, and then fail once, what does that say about it's design?
    Warren> What exactly was fixed in 2.0?  Were the fixes "one liners" and
    Warren> "fencepost errors", or was it a total architecture change?

The code to implement this sort of stuff is not trivial, so it's not
entirely surprising that there are bugs.

Skip Montanaro | http://www.mojam.com/
skip at mojam.com | http://www.musi-cal.com/
"Languages that change by catering to the tastes of non-users tend not to do
so well." - Doug Landauer