------- Comment #2 from pault at gcc dot gnu dot org 2006-05-28 14:09 -------
Created an attachment (id=11524)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=11524&action=view)
Patch for the bug exposed by the fix to this and PR25098
The patch for pr25098 fixes this bug but exposes another.
module integrator
interface
function integrate(f,xmin,xmax)
implicit none
interface
function f(x)
real*8 :: f,x
intent(in) :: x
end function f
end interface
real*8 :: xmin, xmax, integrate
end function integrate
end interface
end module integrator
use integrator
real*8, external :: f
real*8 g
print *,integrate (f,0d0,3d0)
print *,integrate (g,0d0,3d0)
end
gives
[EMAIL PROTECTED] prs]# /svn-4.1/bin/gfortran pr25147.f90
In file pr25147.f90:21
print *,integrate (f,0d0,3d0)
1
Error: Type/rank mismatch in argument 'f' at (1)
ie. It generates an error for OK code and doesn't catch the incorrect code!
The patch for PR25098 corrects the latter but still emits the Type/rank error.
The attached patch combines that for PR25098 and passes the correct code.
I will prepare a testcase and submit tonight.
BTW It regtests on FC5/Athlon1700....
I hope that the thesis is coming along, Tobi?
Cheers
Paul
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25147
