I assume it works now for everyone.

Matt, does this work for you now?

On Sat, Mar 04, 2006 at 05:37:48AM +0000, Bryan O'Sullivan wrote:
> 
> Bryan O'Sullivan <bos@serpentine.com> added the comment:
> 
> Matt said he saw some test failures due to this change, but I've never been able
> to reproduce them.
> 
> Matt, any reason we shouldn't close this as fixed?

Still breaks conistently for me. Something leaves an hg serve running,
causing later tests to fail.

Matt said he saw some test failures due to this change, but I've never been able
to reproduce them.

Matt, any reason we shouldn't close this as fixed?

please look in crew. i have used option name -d,--daemon.
code is little strange, because uses os.spawnvp to be portable.
for msg223, server is completely started when parent process exits. no race.

I don't see how to daemonize and be sure that the server is completely started.
(a classic backgrounding doesn't solve the startup races)

This will help fix startup races with running hg serve in shell scripts.