This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [gfortran] PATCH -- correct underflowing of complex types.


On Sun, Jun 20, 2004 at 08:38:18PM +0200, Tobias Schl?ter wrote:
> Steve Kargl wrote:
> >
> >I don't have commit access.  Can you commit the fix?
> 
> I just checked it in. This looks like one could create a testcase, does 
> something like this trigger an error before your patch?
> 
>    complex, parameter :: c = (3.,3.) - (SQRT(9.),SQRT(9.))
> 
>    if (c.ne.(0.,0.)) call abort()
>    end
> 

I haven't bootstrapped and tested this change because it
seemed obvious.  The code I changed is a cut-n-paste 
bug from an earlier patch I submitted.  The G77 test suite
or NIST suite includes something similar to 

   program b
   implicit none
   real x
   complex y
   x = abs(sqrt(9.125)**2 - 9.125)                                
   y = cabs(cexp((-2.5, 1.375))) - exp(5.0 / (-2.0))
   print *, 'x = ', x
   print *, 'y = ', y
   end program b

The above program didn't test the code path I just changed,
which is why I missed the cut-n-paste bug.

-- 
Steve


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]