Hi Mikael, The regression was fixed by s/rvalue->expr_type == EXPR_NULL/rvalue->expr_type == EXPR_NULL && !rvalue->symtree/
The test is interface_64.f90. Regression tested on FC44/x86_64. I apologise for forgetting about your review - many thanks! Paul On Tue, 20 Oct 2026 at 12:36, Paul Richard Thomas <[email protected]> wrote: > > Hello Mikael, > > I pushed the patch, as it was submitted based on a review by > Jerry/Claude. Pushed as r17-5405. > > I totally forgot about your useful caveat. It will be fixed in the > next 24 hours. Sorry about that. > > Regards > > Paul > > On Tue, 13 Oct 2026 at 13:53, Mikael Morin <[email protected]> wrote: > > > > Le 08/10/2026 à 14:36, Paul Richard Thomas a écrit : > > > The attached patch is trivial, allowing procedure pointer components to > > > be initaialized to NULL (), and fixes bug13-proc-ptr-comp-init-ice.f90. > > > > > > Passes regtesting on FC44/x86_64. OK for mainline. > > > > > > Cheers > > > > > > Paul > > > > > Hello Paul, > > > > the following example is accepted with your patch whereas it is > > (correctly) rejected without it. I think we can pick s2 from > > rvalue->ts.interface if it's set before giving up and returning. > > > > Mikael > > > > program prog > > implicit none > > interface > > integer function foo(arg) > > integer, intent(in) :: arg > > end function > > integer function bar(arg1, arg2) > > integer, intent(in) :: arg1, arg2 > > end function > > end interface > > procedure(foo), pointer :: p > > procedure(bar), pointer :: q > > p => null(q) > > end program > >
