1.1.14+retry segmentation faults seem fixed

I heard from Christian Krackowizer that he was able to apply the retry
patch to 1.1.14 and run it without crashing for the past 40 hours, so just
to make sure, I downloaded a new copy of 1.1.14 and applied the patch to
it.

It looks like the problem was caused, somehow, by the way in which the
timeout value is used in the comm_connect_addr() function. I changed this
after creating the patch in order to get rid of a spurious variable left
over from the setsockopt() approach, thinking that it wouldn't make any
difference. Maybe some kind of type mismatch, or something. *sigh*

The good news is that 1.1.14's retry patch seems to be operational, 3,269
connections here so far over the last half hour.

        -Mike Pelletier.
Received on Thu Jul 31 1997 - 09:31:06 MDT