[Bug fortran/19168] New: Mismatched KINDs in SELECT CASE constucts is not handled correctly

["sgk at troutmask dot apl dot washington dot edu" <gcc-bugzilla@gcc.gnu.org>]

Consider this program posted by James Van Buskirk in c.l.f.
gfortran currently issues an error because the KIND of 7 and
200 do not match the KIND of x.

program case_ex
   implicit none
   integer, parameter :: ik1 = selected_int_kind(2)
   integer(ik1) x
   integer io

   write(*,'(a)',advance='no',iostat=io) ' Enter the value of x:> '
   read(*,*) x
   if(io /= 0) x = 0
   select case(x)
   case(7)
      write(*,'(a)') ' Sorry, bad guess.'
   case(200)
      write(*,'(a)') ' You win the prize!'
   case default
      write(*,'(a)') ' You aren''t even trying.'
   end select
end program case_ex

The relevant constraint from the Fortran standard is 

  C805 (R808) For a given case-construct, each case-value shall be of
  the same type as case-expr.  For character type, length differences
  are allowed, but the kind type parameters shall be the same.  */

and we note no such requirement of the KIND for INTEGER and LOGICAL
variables must match.

In the discussion on c.l.f, Richard Maine pointed out that the 
normal promotion rules for numeric rational operation should be
followed.

An initial patch was posted to http://gcc.gnu.org/ml/fortran/2004-12/msg00179.html
but it was deemed incorrect.

-- 
steve

-- 
           Summary: Mismatched KINDs in SELECT CASE constucts is not handled
                    correctly
           Product: gcc
           Version: 4.0.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: fortran
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: sgk at troutmask dot apl dot washington dot edu
                CC: gcc-bugs at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19168