So I've put M1 on my mail sever (from the branch). Since RC1, kabir has fixed the mail-list-loop bug. Its not running quite as fast as JAMES did but I'm using hypersonic for the database so what do you expect? (I'd expect it to be even slower than it is compared with a file based store). There are things we can do to tune the CMP engine to make pop faster.
Our defaults as mike noted in the descriptor file suck totally. Meaning the "Life" needs to be much much higher and the thread pools should be closer to 100 or so. (I set those low for the purpose of speed tests and forgot to fix them)
I did run into this: http://developer.java.sun.com/developer/bugParade/bugs/4699981.html when the defaults were too low. I haven't seen it since I upped the limits.
We DO need to rationalize our debugging code. I can't seem to find the proper log level to not have every mail repeated in the log! I thought we had this at like TRACE or something. I have a limited amount of disk space left on my mail/web server and this will probably be a problem for me.
All of that being said, these problems are pretty damn minor for a first milestone release. This is better than a lot of people's 1.0. I'll change the defaults to what I think is reasonable for most people and then lets release M1. Please vote.
Everyone with R/W access has a binding vote. We still welcome patch submitters and other "stakeholders" (if you just wandered in then reserve judgement ;-)) to vote but only those with R/W access who have developed on mail services get a binidng vote.
Signify as follows:
+1 = Yes and I will do what I can to help with this and support any fallout
+0 = Its a good idea but I don't care very much (do it, but I won't help you at all)
-0 = You suck, we shouldn't do this but I won't stop you. I think its bad because...
-1 = Whoooaaa, bad idea and here is why...
To release we need at least 2 +1s and no vetos. (When we have more people we'll up this to three but I don't think we should require EVERY active R/W committer to vote). The vote ends when either all active R/Wers have voted or 48 hours has passed.
What the hell ;-)
Very cool. Congratulations. Sorry I've been away, I haven't meant to be. But I got an XBox for my birthday and I've been learning Hibernate at work. I've got my head back up above water now :D
+1 from me as well of course. Note I deleted the impassioned defense of hypersonic as its off topic and took a childish tone. However it did point out one data point I left out: we drastically beat the performance of james on a reasonably modern system with small data sets. We do need more Junit test.
bring em on ---
+1 for the kabir release!
Dudes... I just turned down the logging and this morning tried it out. Turns out it wasn't us/hypersonic or anything. Its sad that I had to turn the logging to like "WARN" or something but now it FLIES. It totally kicks james into the dirt from a performance standpoint. Of course we'll see what happens latter in the day when I get a bazillion emails. I'll roll and annouce the release either today or monday (never release on the weekend or friday because no one sees it ;-) ). I can't wait to see mail lists and IMAP (even though I think I'll need to upgrade my server ;-) ).