[Patch, fortran] PR20779 and PR20891 - allocation of STAT or specification variables in same call

[Steve Kargl]

On Mon, Sep 18, 2006 at 08:12:35PM -0500, Erik Schnetter wrote:
> On Sep 18, 2006, at 17:16:59, Paul Thomas wrote:
> 
> >With or without the patch in place, gfortran (ifort) compile
> >
> >   integer, allocatable :: x(:)
> >   integer :: i, j(2)
> >   allocate (x(kind(i)), stat=i)
> >   print *, size(x)
> >   allocate (x(size(j)), stat=j(1))
> >   print *, size(x)
> > end
> >
> >to give
> >
> >          4
> >          2
> >
> >as they should. ie. size(i) and kind(j) are just perfectly ordinary  
> >expressions.  The patch was concerned to catch the non-standard  
> >case where STAT or a variable in the specification of other  
> >allocated arrays was being allocated in the same statement.
> 
> Paul,
> 
> I think I meant the expression
> 
> 	allocate (i(1), stat = j(kind(i)))
> 
> I got the problem case backwards.  (Not that anybody would write code  
> like that.)  I was wondering whether the code that looks for  
> "variable occurs in expression" handles expressions which do not  
> depend on the value of a variable, and which are therefore allowed.   
> (Or are they?)
> 

I know this is going to sound real crazy, but one can take
patches posted to list, apply to the source code, build
gfortran, and test the result.


program p
  integer j(8)
  integer, allocatable :: i(:)
  allocate(i(1), stat=j(kind(i)))
  print *, j(4)
end program p

gfc4x -o z p.f90
./z
         0


-- 
Steve