I tried the test case below w/the 5/27 snapshot and got
what appear to be the same fork errors.  So, I'm sticking
with the 4/3/2006 snapshot.

Is there any other info I can supply to help?

--
thanks
Tom

On Wed 5/17/06 23:03 CDT cygwin@cygwin.com wrote:
> Thanks to all for trying the test.
--snip
> Here's the nonsense command sequence again:
> 
>   cd /tmp;echo -ne '#!/bin/bash -u\nfoo "$@"\n' >bar;chmod +x bar
>   foo() { for f ;do bash -c true ;done ; } ; export -f foo
>   find /usr -print0|xargs -0 ./bar
> 
> In the function "foo" "f" takes on the value from the find output,
> but absolutely nothing is done with the value "$f".  Maybe the fork
> error would still show up if "find /usr -print0", was replaced w/"yes".

Using 5/27 snapshot,I tried 
  "yes|xargs -0 ./bar" 
but it did not cause fork errors, but "find /usr -print0|xargs -0 ./bar" did.

--snip
> The box that is acting up for me is a Windows 2000 server, w/OOTB heap settings.