Theo Van Dinter wrote: > In my tests, freeze/thaw seems to work nicely, and I've implemented it in > my new spamd code. I haven't quite figured out where it'll fit in an API. > If it goes into the M::SA modules, it'll be a generic requirement, which > I don't want. But leaving it in spamd by itself means that third-party > daemons have to implement it for themselves. :(
Helped along by Theos spamd changes, I've implemented the proposed freeze thaw in spampd and it works here too. Seems to use quite a bit of memory though, looks like 4-5M. I have yet to study the performance-impact in thawing a Conf object, but given that I really need the correct per-user behaviour, it's not overly important. Many thanks for the help, guys! /Per Jessen, Zurich
