From: cgf AT cygnus DOT com (Christopher Faylor)
Subject: Re: error when maxing out processes
3 Jun 1998 19:36:15 -0700
Message-ID: <199806040228.WAA12994.cygnus.cygwin32.developers@kramden.cygnus.com>
To: newsham AT lava DOT net
Cc: cygwin32-developers AT cygnus DOT com

>> >I noticed that the code has a loop over each child doing WaitForSingleObject
>> >in the event that the WaitForMultipleObjects failed.  Does this mean
>> >the code already properly handles this limitation (in some degraded way)?
>> 
>> That loop is to detect errors.  If WaitForMultipleObjects detects an
>> error, that loop should pinpoint which entry in the array is in error.
>> Theoretically, it shouldn't be necessary but someone on the mailing list
>> just reported an error there today.  Sigh.  
>
>So the wait_subproc really doesnt work when there are more than
>64 children?  So what is the end result that we can expect to see
>from this as application programmers?

I don't know that it has ever been tested.  If I had to theorize I
would guess that you might 1) get the error you reported, 2) you might
see some processes exiting but not noticed immediately or, 3) you
might get an immediate error when the number of handles being waited
for is >64.  Actually, 3 is the same as 1.

So, I would guess that at the 64th fork or spawn you'd get exactly the
error you're seeing.

cgf