From Bugzilla Helper:
User-Agent: Mozilla/5.0 Galeon/1.2.5 (X11; Linux i686; U;) Gecko/20021003
Description of problem:
ORBit assumes the existence of a bug in getpeername, which is no longer present
in 2.5 kernels.
This causes Evolution to malfunction.
The patch is a one-line change and is contained in this message to the
evolution-hackers mailing list:
This patch solves the problem on my system: please apply.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Boot Linux 2.5.49 (any other recent 2.5 kernel should also cause the problem)
2. Start Evolution (which uses ORBit)
3. Try to send a message
Actual Results: Evolution hangs.
Expected Results: The message is sent.
I wouldn't consider this a bug considering Red Hat does not release any of the
2.5 development kernels. It should be up to the user to patch everything on
his/her system to accomodate a newer kernel.
Otherwise - all of us people who choose a vanilla Red Hat system would be
affected by the patch as well (could cause regression problems n' such).
Since the 2.6 kernel will relatively be new by the time the new Red Hat
distribution is released (if the 2.6 kernel is even released by that point),
even filing this bug as an enhancement seems trivial at best.
Well, it's good to know about this anyhow.
I wonder how the heck orbit manages to get confused
by this change...
The bug turned out to be a kernel bug, fixed in 2.5.55: when getpeername was
called on a unix socket, the kernel returned the name of the socket itself
rather than the one of the peer.
Please don't apply the workaround patch.