This is the mail archive of the
mailing list for the Cygwin project.
Re: simple test triggers fork errs for me in 5/27 snapshot
On Mon, May 29, 2006 at 11:38:18AM -0400, Christopher Faylor wrote:
>On Mon, May 29, 2006 at 02:03:43PM +0100, Dave Korn wrote:
>>Anyone got any good ideas?
>I've rescanned this thread to reacquaint myself with why I was
>semi-ignoring it and I've found that the data points are 1) it fails
>with 4/27 and one subsequent snapshots 2) it works with the 4/3
>snapshot 3) it fails with the 3/9 snapshot.
>I was hoping that somebody besides me would offer the obvious
>observation that if this truly stopped working at some point, then
>knowing when it stopped working with less granularity than a 24 day
>period would be the first step towards fixing the problem.
>So, that's my good idea.
I couldn't duplicate this on my main multithreaded windows system (which
is one of the reasons why I was hoping for someone else to step in and
help with the more obvious aspects of this debugging exercise) but I
could duplicate it on the "game computer". I discovered that the first
failing snapshot was 4/12. And, of course, that's the snapshot where I
put the windows command-line stuff back in for cygwin processes.
However, when deciding how small to make the windows command line
buffer, I chose too large a value for filling out the windows command
line for cygwin processes.
I've reduced the size of the windows command line which is sent to
Cygwin processes still further and that fixes the problem on my game
computer. That fact that this worked just fine on my normal test
computer (and presumably on Corinna's computer) is why this problem
crept back in. So, that leads me to believe that this problem is
system-dependent. I hope the constant that I chose is small enough
for all of the systems out there.
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
Problem reports: http://cygwin.com/problems.html