Raymond Toy wrote:
> >>>>> "Raymond" == Raymond Toy <[EMAIL PROTECTED]> writes:
>
> Raymond> William Brew wrote:
> >> Oops, I included the transcript twice. Attached is the
> >> lisp code.
>
> Raymond> Thanks for the code. I can reproduce the problem.
>
> Raymond> And, more importantly, I think I know what's causing it. I think the
> Raymond> default stack on Solaris is 8 MB. That is not enough. (parse-programs
> Raymond> 6 1500 1000) will cause a core dump. If I up the stack to 16 MB, it
> Raymond> purifies without problem.
>
> Raymond> I guess you are creating something that is deeply nested and pscav is
> Raymond> calling itself recursively many, many times and the stack is exhausted.
>
> FWIW, I measured this, and there are at least 17000 nested calls to
> pscav. A stack frame is at least 64 bytes, and each call to pscav
> from pscav has a couple of other calls, so we're talking at least 3 MB
> of stack.
>
> So does increasing the stack help? Or does your real application
> exhibit some other problem now? It would be nice to know.
Q: So does increasing the stack help?
A: I'm confused. Your previous assertion was that a larger stack appeared
to solve the problem. The 3M estimate seems to say that the existing 8M
stack was plenty big.
Q: Does the real application exhibit some other problem now?
A: Is there any way for me to test the real application with a larger stack other than
to get a new snapshot? (if there is, then I don't know what it is) If not, will the
forth coming snapshot have a larger stack for Solaris?
-bill-
