bug#13546: 24.2.92; Error(s) when sending emails

[Eli Zaretskii <eliz@gnu.org>]

> From: "Sebastien Vauban" <wxhgmqzgwmuf@spammotel.com>
> Date: Fri, 15 Feb 2013 21:50:27 +0100
> Cc: 13546@debbugs.gnu.org
> 
> >> - emacs-24-r111270-20130215-w32-i386.zip (58.94 MB), 1 hr ago
> >
> > That one please.
> 
> Check your inbox, I've sent you a screencast link (which I prefer to keep
> private, for not exposing private stuff), but please know that version does
> not work at all for my test case:
> 
> - using Helm: buffers shown, locate not displayed...

Any error messages?  Anything in *Messages*?

> - creating or opening an Org file (I wanted to test the export but couldn't
>   get that far: max-lisp-eval-depth problem!

That's got to be unrelated.  Does it help to enlarge
max-lisp-eval-depth?

> - reading my emails with Gnus: server is inaccessible

Is that right from the start?  If so, can you open any other network
connection, like fetch some URL with the url-retrieve?

Can you launch a process?  E.g., does this work, if evaluated in
*scratch*?

  (setq proc (start-process-shell-command
		"locate<0>" "locate-results<0>"
		"locate eee"))
  (sleep-for 20)
  (delete-process proc)

If you attach GDB, do you see any "warning" messages?

I'm at a loss here...  The changes don't affect any code that launches
subprocesses or open network streams.  They just affect what happens
when there are no more free slots for additional subprocesses, which
should only happen once you've launched at least 32 of them.  Can you
show the contents of child_procs[] array, like you did before, when
you cannot see the results of 'locate'?