> > It would have been a lot easier to define our own rules format, but Sieve >is an existing standard so we went with it, and libSieve was a lot faster >than trying to write an entire parser. > > Aaron's a real nice guy, so if you want to propose changes or send in >patches, I'm sure he'd be very happy to accomodate. >
Thats true. The biggest problem I had was tracking through the levels of abstraction of the interface to determine why certain addresses wouldn't match. I did find that the grammar couldn't match on just a name (local user only) so we attach the node name which might not be correct depending on your Citadel setup. Also there was something with putting another MTA in front of Citadel that caused the addresses to get re-written into a local format that might not match. I think it was Hans that had the most problems with Sieve regarding the "vacation" feature. IIRC he eventually removed postfix (or whatever) from in front of Citadel so that the correct addresses got passed (un molested) to Citadel. It was a nightmare, for me, working out what was going on. I was also new to Citadel back then and didn't fully understand how Citadel stored messages and their headers which didn't help.
