Re: NAN and -INF: precision problems when exchanging between plataforms?

From: Paula (pcoelho_at_usp.br)
Date: 03/22/04 

Date: 22 Mar 2004 12:42:21 -0800

Toon Moene <toon@moene.indiv.nluug.nl> wrote in message news:<405b5791$0$280$4d4ebb8e@news.nl.uu.net>...

> Specifically:
>
> -fno-automatic -finit-local-zero
>
> Hope this helps,

Hi, here I am again.

So, if I use -fno-automatic while compiling g77 it works!! The option
-finit-local-zero doesn't change the behaviour of the code, probably
because I already had identified the non initialized variables and
correct this. One less non-conformance at least....

If I compile in Tru64 with -recursive option, I got:

sendsig: can't grow stack, pid 2749, proc p, sig 11
Segmentation fault

just when I start the program.

So, I'm happy that the code is running. I'll try to find the source of
the problem with the trap code by Tom Crane. I didn't understood the
error message in Tru64, but concerning the behaviour in linux, as far
as I understood from what you all wrote, probably the problem is in a
missing SAVE statement in a subroutine that is called several times,
is this?

Cheers,
Paula.