Re: floating point exceptions: Linux kernel feature or bug?

[Michal Jaegermann <michal ellpspace math ualberta ca>]

> 
> Thanks for any help and feel free to e-mail me directly rather than
> via the newsgroups and/or mailing lists.
> 
> test.f:
> ___________________________________________________________
>       program test
> c
> c     program to test g77 on i386 and alpha
> c
>       x=1.0
>       y=0.0
>       z=2.0
> 
>       a=x/z
>       write(*,*)x,z,a
> 
>       b=x/y
>       write(*,*)x,y,b
> 
>       end
> ____________________________________________________________

This program compiled on Alpha with 'g77 -O2 -mieee -o check test.f'
produces the following output:

  1.  2.  0.5
  1.  0.        Inf.

g77 from egcs-1.0.2 release.  Options -mieee, -mieee-with-inexact
and -mieee-conformant are documented in gcc info files so one has
to know, I am afraid, that g77 will accept gcc flags on the top of
its own specific stuff.

BTW - never call your test programs 'test' or you may be for a nasty
surprise. :-)  If you want to know why type, in bash, 'which test'. :-)

 Michal