Hi. I have been unable to send an annnouncement for b10 due to lack of
internet connectivity. That version turned out to be buggy anyway; if you
have downloaded it by accident, please bin it.

Anyway, I have just released b11, which has many improvements. Mathopd can
now use vfork() to spawn child processes for CGI. This turns out to yield
a dramatic speed improvement. On my platform (FreeBSD) about 50%. I will
try to do some exact benchmarks later. Another improvement is that Mathopd
will no longer loop over all inactive connections in its main loop. The
difference in speed will be noticeable especially if NumConnections is set
to a very large number.

If you want to have the benefits of vfork, you must compile with
-DHAVE_VFORK. See the Makefile for details.

Cheers
Michiel

Reply via email to