Sidney Markowitz writes: > Justin Mason wrote, On 24/4/07 12:24 AM: > > Hey all -- I think the win32 stuff is looking good now, so here's an > > rc3 proposal. Shout if you think it's premature, but I think we're > > ready for it. Hopefully this'll be the last rc, then we can put > > out 3.2.0 as a full release... > > There is one more issue I see in both win32 and cygwin: I'm getting > failures in dnsbl.t and dnsbl_sc_meta.t which I've tracked down to only > one of all the backgrounded queries coming back with a reply. When I do > a simple spamassassin -D dns -t < sample-nonspam.txt I see no background > queries returning with the default rbl_timeout. When I increase it to 60 > as in the dns tests, I get a variable number returning between 2 and 15 > or so. > > I thought this might have something to do with the high latency or > perhaps packet loss on my satellite Internet connection, but I see no > such problems running under Mac OS 10.4.9. > > Can anyone please see if the dnsbl tests work at all under win32 or > cygwin? Is this enough to block the rc release?
They all pass for me on win32... I'm afraid it may be your high-latency setup :( > By the way, in checking this out I noticed that AsyncLoop.pm has a > timeout after a hardcoded 20 one-second repetitions of no change in the > number of background queries. That was causing the tests I was doing to > time out after 20 seconds rather than the 60 I had set rbl_timeout to. > Do we want that hardcoded or should it be a config option? hmm. that probably needs to be derived from rbl_timeout somehow. --j.
