[Bug fortran/69739] [4.9/5/6 Regression] ICE during array result, allocatable assignment

[anlauf at gmx dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69739

Harald Anlauf <anlauf at gmx dot de> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |anlauf at gmx dot de

--- Comment #3 from Harald Anlauf <anlauf at gmx dot de> ---
(In reply to John from comment #2)
> After reporting the bug, I noticed that in the actual code I have (not the
> reduced version I posted before), adding the (:) stride doesn't always make
> the ICE go away.
> 
> What solved the problem once and for all (and without using the
> -fno-realloc-lhs flag) was changing the result in the "operate" function to
> be that of an allocatable array:
> 
>     function operate(A, X)
>         type(sometype), intent(IN) :: A(:,:,:)
>         real, intent(IN) :: X(:)
> !        real :: operate(1:PRODUCT(UBOUND(A)))
>         real, allocatable :: operate(:)
> 
>         allocate (operate(1:PRODUCT(UBOUND(A))))
>         operate(:) = 0
>     end function
> 
> So maybe the automatic-size result is the actual root of the ICE?

If I use

        real              :: operate(1:size(A))

instead of

        real              :: operate(1:PRODUCT(UBOUND(A)))

I do not get an ICE.  So it's not automatic-size per se.