g77 testsuite (was RE: IMPLICIT CHARACTER*# causing 35 regressions in NIST testsuite)

Steve Kargl sgk@troutmask.apl.washington.edu
Thu Jul 8 01:50:00 GMT 2004


On Thu, Jul 08, 2004 at 01:14:27AM +0100, Paul Brook wrote:
> > One area of failures in the G77TS are the bit manipulation functions
> > (eg, IBSET).  gfortran only supports these functions with default
> > integer kind while the G77TS has tests for INTEGER*1, INTEGER*2, and
> > INTEGER*8.
> 
> I wasn't aware of this. The backend implementations in trans-intrinsic.c 
> should work with any integer kind.
> 

Here just a sampling of the failures.

 In file f90-intrinsic-bit.f:143

      call c_i1(IBSET(k,k2),ka,'IBSET(integer*1,integer*1)')
                       1
Error: 'pos' argument of 'ibset' intrinsic at (1) must be of kind 4
 In file f90-intrinsic-bit.f:144

      call c_i1(IBSET(k,m2),ka,'IBSET(integer*1,integer*8)')
                       1
Error: 'pos' argument of 'ibset' intrinsic at (1) must be of kind 4
 In file f90-intrinsic-bit.f:146

      call c_i8(IBSET(m,j2),ma,'IBSET(integer*8,integer*2)')
                       1
Error: 'pos' argument of 'ibset' intrinsic at (1) must be of kind 4
 In file f90-intrinsic-bit.f:147

      call c_i8(IBSET(m,k2),ma,'IBSET(integer*8,integer*1)')
                       1
Error: 'pos' argument of 'ibset' intrinsic at (1) must be of kind 4
 In file f90-intrinsic-bit.f:148

      call c_i8(IBSET(m,m2),ma,'IBSET(integer*8,integer*8)')
                       1
Error: 'pos' argument of 'ibset' intrinsic at (1) must be of kind 4


-- 
Steve



More information about the Fortran mailing list