On Wed, Apr 25, 2007 at 12:07:19AM +0200, Gilles LAMIRAL wrote: > > I think it should just check and if imapd doesn't know MD5 auth just don't > > use it, unless explicitly told to do so. > > This used to be the behavior. There was a time imapsync exited > in than case but I received two many complains. > > We also found servers not claiming they support some AUTH > mechanism but actually supporting it. imapsync tries with
ok, I see the point. > the problem here is the 99.8% cpu not stopping. I wonder > why, what says strace ? write(1, "Host mailhost says it has NO CAPABI"..., 63Host comix says it has NO CAPABILITY for AUTHENTICATE CRAM-MD5 ) = 63 write(3, "\27\3\1\0000z\210e\263\27\361\347\356f\24\310\235W\264"..., 53) = 53 read(3, "\27\3\1\0H", 5) = 5 read(3, "5\n\375\250\247]\351^\240\344dn\206~\340\336\26O,\215\32"..., 72) = 72 read(3, 0x85fa570, 5) = -1 EAGAIN (Resource temporarily unavailable) read(3, 0x85fa570, 5) = -1 EAGAIN (Resource temporarily unavailable) ... fd 3 is the socket. > There are several tricks to improve memory and speed. > --useheader 'Message-ID' > is one. ok, got it - it's ~like with mailsync(1) - should work, I've ensured fetched email has proper msg-id since quite a bit by now. In the past mailsync failed sometimes because of msgs with no msg-id. > In fact, for the main .deb port, the difficulty seems not > technical, but social, having the keys to do it without > knowing anyone in the debian team may take some time and > energy. yep, with such a large sw and devel/contrib collection to manage, a strong policy is mandatory. > In Debian every change seems a heavy work. I don't know why the point is that once a released, a 'stable' Debian is frozen wrt features. So bugfix from upstream need to be backported, leaving new features to 'unstable'. Also, steps needed for any patch to reach next .deb release should ensure some scrutiny by QA & security team, or at least more eyes than .deb maintaner's only. That's the hope, that is. > official software releases. For imapsync it's just 8 months, > a good point. :) thanks -- paolo -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
