[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

970520: netcdf 3.3 with gcc

Kate,

>Date: Tue, 20 May 1997 11:05:22 -0400 (EDT) 
>From: Kate Hedstrom <address@hidden>
>Organization: Rutgers University
>To: address@hidden
>Subject: Re: 970520: netcdf 3.3 with gcc 
>Keywords: 199705171514.JAA09636

In the above message, you wrote:

[Talking about RAND_MAX]
> I wasn't sure which it should be and tried both.  It didn't seem to
> make any difference.

It won't make any difference to the correct operation of the testing
routine because rand(3) is only used to arbitrarily chose one of several
(hopefully correct) execution paths (to reduce time) -- and any path
should be correct.  What will happen is that the decision will be biased
if the wrong RAND_MAX is used -- resulting in still correct behavior but
less coverage.

> > How slow is "unbelievably" slow?
> 
> Well, this time it took:
> 
> 35.820u 7.620s 1:03.72 68.1% 0+428k 224+490io 220pf+0w
> 
> Last week it took over an hour - I don't know what changed.

Whatever it was, it was for the better.  :-)

> 
> > > I have
> > > tried to figure out why nf_test is dumping core, but have not
> > > succeeded.  I don't expect you to figure this out, but wanted you to
> > > know about the gcc changes.
> > 
> > I believe the reason it crashes has to do with the way the FORTRAN
> > runtime is (mis)handling the value conversion errors (e.g. converting an
> > out-of-range double into a signed byte).  I'll investigate.
> 
> gcc on FreeBSD failed (dumped core) on the cast in:
> 
> #include <stdio.h>
> 
> main() {
>     double d = -2147483649.;
>     int i;
>     i = (int)d;
>     printf("%d  %lf\n", i, d);
> }

Tsk, tsk!

> 
> I reported it to the gnu people.

Thanks!

--------
Steve Emmerson   <address@hidden>