------- Additional Comments From tobi at gcc dot gnu dot org 2005-07-09 12:04
-------
The real bug is in gfc_conv_expr_descriptor, as proven by the following
testcase, which doesn't exercise the codepath you're modifying, but the parts
Paul's patch makes us evade:
[EMAIL PROTECTED] tests]$ cat derivedptr.f90
type a
integer :: i, j
end type a
type(a), target :: t(5)
integer, pointer :: p(:)
t(:)%i = (/1,2,3,4,5/)
print *, t(:)%i
p => t(:)%i
print *, p
end
[EMAIL PROTECTED] tests]$ ~/src/gcc-new/build/gcc/f951 derivedptr.f90 -quiet
[EMAIL PROTECTED] tests]$ gfortran derivedptr.s
[EMAIL PROTECTED] tests]$ ./a.out
1 2 3 4 5
1 11507280 2 134520840 3
[EMAIL PROTECTED] tests]$
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18022
