some problems with gcc/testsuite/gfortran.dg/ tests in 4.3-20061111 with OSX 10.3

[Dominique Dhumieres]

> The code for sub is:
> 
> sub (__result, .__result, i)
> {
> ...
> }
> 
> which looks OK apart from the repetition of ..__result = *i;  However, I 
> can see nothing that would make the program hang - can anybody else?

I get the same code, excepted the varaible labels that have a -44 offset 
(int4 D.1001; instead of int4 D.1045; and so on).

The problem I see is with "recursive" WRITE as in:

integer j
j = f(10)
print *, j
print *, f(10)
contains
integer function f(i)
  integer i
  print *, i
  f = 2*i
end function f
end

where the executable hangs at the 'print *, f(10)' level.
I wonder if this is a general problem with OSX on PPC or if this
specific to the dirty hack I have done to build the 20061111 snapshot
on OSX 10.3 (c, c++, fortran, objc, and java have been successfully built 
and all the other tests I have done with gfortran pass, including
gcc/testsuite/gfortran.dg/write_recursive.f90).

Dominique