[CDBI] Re: construct sucks.

Perrin Harkins wrote:
> The potential problem with this change (and I think this was actually
> discussed earlier on the list) is that somewhere else you may have
> already updated one of those keys and not saved it yet, so you'd be
> overwriting that update here. You could check to see if the keys have
> been changed before writing the new data to the object, although this
> will slow it down.

If this problem (and any others that might crop up) were solved, however, it
does seem like it could be a very useful addition, even with the extra
checking. More experimentation and a test case (if there isn't already one)
that properly touches on the potential problem cited by Perrin above are
needed.