https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121616
kargls at comcast dot net changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |kargls at comcast dot net
--- Comment #3 from kargls at comcast dot net ---
(In reply to Jean Vézina from comment #0)
> Created attachment 62161 [details]
> Sample program to reproduce the problem
>
> Assignment of a value of intrinsic type (real, integer, etc) to an unlimited
> polymorphic variable is not done correctly because a segmentation fault
> occurs when a reference to the variable is made after.
It is sometimes correctly assigned. :-)
It seems that there may be an issue involving the resolution of an
unlimited polymorphic component when a left part-ref is an array.
% cat tr.F90
program foo
type general
class(*), allocatable :: x
end type
#ifdef WORKS
type(general) a
a%x = 1
call prn(a%x)
#else
type(general) a(1)
a(1)%x = 1
call prn(a(1)%x)
#endif
contains
subroutine prn(x)
class(*), intent(in) :: x
select type(x)
type is(integer)
stop 1
type is(real)
stop 2
end select
end subroutine
end program foo
% gfcx -o z tr.F90 -DWORKS && ./z
STOP 1
% gfcx -o z tr.F90 && ./z
Segmentation fault (core dumped)
