On Fri, Jul 26, 2024 at 09:49:27PM +0200, Jakub Jelinek wrote: > On Fri, Jul 26, 2024 at 08:43:44PM +0200, Jakub Jelinek wrote: > > Yeah, I saw ARGUMENT_PACK_SELECT being used, but didn't notice that > > important > > if (arg_pack && TREE_CODE (arg_pack) == ARGUMENT_PACK_SELECT) > > arg_pack = ARGUMENT_PACK_SELECT_FROM_PACK (arg_pack); > > part of tsubst_pack_expansion where it picks the pack from > > ARGUMENT_PACK_SELECT when needed. > > > > I also saw the iterative_hash_template_arg case and was afraid to modify it > > because of that reuse of ARGUMENT_PACK_SELECT tree. > > But I guess you're right, if there is a flag added to it only set by > > satisfy_fold which needs to create them immutable because there is the > > satisfaction cache etc., it might work. Will try that soon. > > With following incremental patch I've made some progress, but still > something to debug...
One thing which needs solving is // P2963R3 - Ordering of constraints involving fold expressions // { dg-do compile { target c++20 } } template <class ...T> concept C = (__is_same (T, int) && ...); template <typename V> struct S { template <class ...U> requires (C<U...>) static constexpr bool foo () { return true; } }; static_assert (S<void>::foo <int, int, int, int> ()); somehow the template parameter mapping needs to be remembered even for the fold expanded constraint, right now the patch will see the pack is T, which is level 1 index 0, but args aren't arguments of the C concept, but of the foo function template. One can also use requires (C<int, int, int>) etc., no? Jakub