* [PATCH] c++: CTAD from initializer list [PR106366]
@ 2022-07-21 12:43 Patrick Palka
2022-07-21 17:06 ` Patrick Palka
2022-07-22 21:32 ` Jason Merrill
0 siblings, 2 replies; 3+ messages in thread
From: Patrick Palka @ 2022-07-21 12:43 UTC (permalink / raw)
To: gcc-patches
During CTAD, we currently perform the first phase of overload resolution
from [over.match.list] only if the class template has a list constructor.
But according to [over.match.class.deduct]/4 it should be enough to just
have a guide that looks like a list constructor (which is a more general
criterion in light of user-defined guides).
Bootstrapped and regtested on x86_64-pc-linux-gnu, does this look OK for
trunk?
PR c++/106366
gcc/cp/ChangeLog:
* pt.cc (do_class_deduction): Don't consider TYPE_HAS_LIST_CTOR
when setting try_list_ctor. Reset args even when try_list_ctor
is true and there are no list candidates. Call resolve_args on
the reset args.
gcc/testsuite/ChangeLog:
* g++.dg/cpp1z/class-deduction112.C: New test.
---
gcc/cp/pt.cc | 25 +++++++++----------
.../g++.dg/cpp1z/class-deduction112.C | 14 +++++++++++
2 files changed, 26 insertions(+), 13 deletions(-)
create mode 100644 gcc/testsuite/g++.dg/cpp1z/class-deduction112.C
diff --git a/gcc/cp/pt.cc b/gcc/cp/pt.cc
index 718dfa5bfa8..0f26d6f5bce 100644
--- a/gcc/cp/pt.cc
+++ b/gcc/cp/pt.cc
@@ -30250,8 +30250,8 @@ do_class_deduction (tree ptype, tree tmpl, tree init,
else if (BRACE_ENCLOSED_INITIALIZER_P (init))
{
list_init_p = true;
- try_list_ctor = TYPE_HAS_LIST_CTOR (type);
- if (try_list_ctor && CONSTRUCTOR_NELTS (init) == 1
+ try_list_ctor = true;
+ if (CONSTRUCTOR_NELTS (init) == 1
&& !CONSTRUCTOR_IS_DESIGNATED_INIT (init))
{
/* As an exception, the first phase in 16.3.1.7 (considering the
@@ -30310,26 +30310,25 @@ do_class_deduction (tree ptype, tree tmpl, tree init,
tree fndecl = error_mark_node;
- /* If this is list-initialization and the class has a list constructor, first
+ /* If this is list-initialization and the class has a list guide, first
try deducing from the list as a single argument, as [over.match.list]. */
- tree list_cands = NULL_TREE;
- if (try_list_ctor && cands)
- for (lkp_iterator iter (cands); iter; ++iter)
- {
- tree dg = *iter;
+ if (try_list_ctor)
+ {
+ tree list_cands = NULL_TREE;
+ for (tree dg : lkp_range (cands))
if (is_list_ctor (dg))
list_cands = lookup_add (dg, list_cands);
- }
- if (list_cands)
- {
- fndecl = perform_dguide_overload_resolution (list_cands, args, tf_none);
-
+ if (list_cands)
+ fndecl = perform_dguide_overload_resolution (list_cands, args, tf_none);
if (fndecl == error_mark_node)
{
/* That didn't work, now try treating the list as a sequence of
arguments. */
release_tree_vector (args);
args = make_tree_vector_from_ctor (init);
+ args = resolve_args (args, complain);
+ if (args == NULL)
+ return error_mark_node;
}
}
diff --git a/gcc/testsuite/g++.dg/cpp1z/class-deduction112.C b/gcc/testsuite/g++.dg/cpp1z/class-deduction112.C
new file mode 100644
index 00000000000..8da5868ff98
--- /dev/null
+++ b/gcc/testsuite/g++.dg/cpp1z/class-deduction112.C
@@ -0,0 +1,14 @@
+// PR c++/106366
+// { dg-do compile { target c++17 } }
+
+#include <initializer_list>
+
+template<class T>
+struct A { A(...); };
+
+template<typename T>
+A(std::initializer_list<T>) -> A<T>;
+
+A a{1,2,3};
+using type = decltype(a);
+using type = A<int>;
--
2.37.1.208.ge72d93e88c
^ permalink raw reply [flat|nested] 3+ messages in thread* Re: [PATCH] c++: CTAD from initializer list [PR106366]
2022-07-21 12:43 [PATCH] c++: CTAD from initializer list [PR106366] Patrick Palka
@ 2022-07-21 17:06 ` Patrick Palka
2022-07-22 21:32 ` Jason Merrill
1 sibling, 0 replies; 3+ messages in thread
From: Patrick Palka @ 2022-07-21 17:06 UTC (permalink / raw)
To: Patrick Palka; +Cc: gcc-patches, jason
On Thu, 21 Jul 2022, Patrick Palka wrote:
> During CTAD, we currently perform the first phase of overload resolution
> from [over.match.list] only if the class template has a list constructor.
> But according to [over.match.class.deduct]/4 it should be enough to just
> have a guide that looks like a list constructor (which is a more general
> criterion in light of user-defined guides).
>
> Bootstrapped and regtested on x86_64-pc-linux-gnu, does this look OK for
> trunk?
>
> PR c++/106366
>
> gcc/cp/ChangeLog:
>
> * pt.cc (do_class_deduction): Don't consider TYPE_HAS_LIST_CTOR
> when setting try_list_ctor. Reset args even when try_list_ctor
> is true and there are no list candidates. Call resolve_args on
> the reset args.
>
> gcc/testsuite/ChangeLog:
>
> * g++.dg/cpp1z/class-deduction112.C: New test.
> ---
> gcc/cp/pt.cc | 25 +++++++++----------
> .../g++.dg/cpp1z/class-deduction112.C | 14 +++++++++++
> 2 files changed, 26 insertions(+), 13 deletions(-)
> create mode 100644 gcc/testsuite/g++.dg/cpp1z/class-deduction112.C
>
> diff --git a/gcc/cp/pt.cc b/gcc/cp/pt.cc
> index 718dfa5bfa8..0f26d6f5bce 100644
> --- a/gcc/cp/pt.cc
> +++ b/gcc/cp/pt.cc
> @@ -30250,8 +30250,8 @@ do_class_deduction (tree ptype, tree tmpl, tree init,
> else if (BRACE_ENCLOSED_INITIALIZER_P (init))
> {
> list_init_p = true;
> - try_list_ctor = TYPE_HAS_LIST_CTOR (type);
> - if (try_list_ctor && CONSTRUCTOR_NELTS (init) == 1
> + try_list_ctor = true;
I suppose try_list_cand would be a more appropriate name for this
variable now, consider that fixed.
> + if (CONSTRUCTOR_NELTS (init) == 1
> && !CONSTRUCTOR_IS_DESIGNATED_INIT (init))
> {
> /* As an exception, the first phase in 16.3.1.7 (considering the
> @@ -30310,26 +30310,25 @@ do_class_deduction (tree ptype, tree tmpl, tree init,
>
> tree fndecl = error_mark_node;
>
> - /* If this is list-initialization and the class has a list constructor, first
> + /* If this is list-initialization and the class has a list guide, first
> try deducing from the list as a single argument, as [over.match.list]. */
> - tree list_cands = NULL_TREE;
> - if (try_list_ctor && cands)
> - for (lkp_iterator iter (cands); iter; ++iter)
> - {
> - tree dg = *iter;
> + if (try_list_ctor)
> + {
> + tree list_cands = NULL_TREE;
> + for (tree dg : lkp_range (cands))
> if (is_list_ctor (dg))
> list_cands = lookup_add (dg, list_cands);
> - }
> - if (list_cands)
> - {
> - fndecl = perform_dguide_overload_resolution (list_cands, args, tf_none);
> -
> + if (list_cands)
> + fndecl = perform_dguide_overload_resolution (list_cands, args, tf_none);
> if (fndecl == error_mark_node)
> {
> /* That didn't work, now try treating the list as a sequence of
> arguments. */
> release_tree_vector (args);
> args = make_tree_vector_from_ctor (init);
> + args = resolve_args (args, complain);
> + if (args == NULL)
> + return error_mark_node;
> }
> }
>
> diff --git a/gcc/testsuite/g++.dg/cpp1z/class-deduction112.C b/gcc/testsuite/g++.dg/cpp1z/class-deduction112.C
> new file mode 100644
> index 00000000000..8da5868ff98
> --- /dev/null
> +++ b/gcc/testsuite/g++.dg/cpp1z/class-deduction112.C
> @@ -0,0 +1,14 @@
> +// PR c++/106366
> +// { dg-do compile { target c++17 } }
> +
> +#include <initializer_list>
> +
> +template<class T>
> +struct A { A(...); };
> +
> +template<typename T>
> +A(std::initializer_list<T>) -> A<T>;
> +
> +A a{1,2,3};
> +using type = decltype(a);
> +using type = A<int>;
> --
> 2.37.1.208.ge72d93e88c
>
>
^ permalink raw reply [flat|nested] 3+ messages in thread* Re: [PATCH] c++: CTAD from initializer list [PR106366]
2022-07-21 12:43 [PATCH] c++: CTAD from initializer list [PR106366] Patrick Palka
2022-07-21 17:06 ` Patrick Palka
@ 2022-07-22 21:32 ` Jason Merrill
1 sibling, 0 replies; 3+ messages in thread
From: Jason Merrill @ 2022-07-22 21:32 UTC (permalink / raw)
To: Patrick Palka, gcc-patches
On 7/21/22 08:43, Patrick Palka wrote:
> During CTAD, we currently perform the first phase of overload resolution
> from [over.match.list] only if the class template has a list constructor.
> But according to [over.match.class.deduct]/4 it should be enough to just
> have a guide that looks like a list constructor (which is a more general
> criterion in light of user-defined guides).
>
> Bootstrapped and regtested on x86_64-pc-linux-gnu, does this look OK for
> trunk?
OK.
> PR c++/106366
>
> gcc/cp/ChangeLog:
>
> * pt.cc (do_class_deduction): Don't consider TYPE_HAS_LIST_CTOR
> when setting try_list_ctor. Reset args even when try_list_ctor
> is true and there are no list candidates. Call resolve_args on
> the reset args.
>
> gcc/testsuite/ChangeLog:
>
> * g++.dg/cpp1z/class-deduction112.C: New test.
> ---
> gcc/cp/pt.cc | 25 +++++++++----------
> .../g++.dg/cpp1z/class-deduction112.C | 14 +++++++++++
> 2 files changed, 26 insertions(+), 13 deletions(-)
> create mode 100644 gcc/testsuite/g++.dg/cpp1z/class-deduction112.C
>
> diff --git a/gcc/cp/pt.cc b/gcc/cp/pt.cc
> index 718dfa5bfa8..0f26d6f5bce 100644
> --- a/gcc/cp/pt.cc
> +++ b/gcc/cp/pt.cc
> @@ -30250,8 +30250,8 @@ do_class_deduction (tree ptype, tree tmpl, tree init,
> else if (BRACE_ENCLOSED_INITIALIZER_P (init))
> {
> list_init_p = true;
> - try_list_ctor = TYPE_HAS_LIST_CTOR (type);
> - if (try_list_ctor && CONSTRUCTOR_NELTS (init) == 1
> + try_list_ctor = true;
> + if (CONSTRUCTOR_NELTS (init) == 1
> && !CONSTRUCTOR_IS_DESIGNATED_INIT (init))
> {
> /* As an exception, the first phase in 16.3.1.7 (considering the
> @@ -30310,26 +30310,25 @@ do_class_deduction (tree ptype, tree tmpl, tree init,
>
> tree fndecl = error_mark_node;
>
> - /* If this is list-initialization and the class has a list constructor, first
> + /* If this is list-initialization and the class has a list guide, first
> try deducing from the list as a single argument, as [over.match.list]. */
> - tree list_cands = NULL_TREE;
> - if (try_list_ctor && cands)
> - for (lkp_iterator iter (cands); iter; ++iter)
> - {
> - tree dg = *iter;
> + if (try_list_ctor)
> + {
> + tree list_cands = NULL_TREE;
> + for (tree dg : lkp_range (cands))
> if (is_list_ctor (dg))
> list_cands = lookup_add (dg, list_cands);
> - }
> - if (list_cands)
> - {
> - fndecl = perform_dguide_overload_resolution (list_cands, args, tf_none);
> -
> + if (list_cands)
> + fndecl = perform_dguide_overload_resolution (list_cands, args, tf_none);
> if (fndecl == error_mark_node)
> {
> /* That didn't work, now try treating the list as a sequence of
> arguments. */
> release_tree_vector (args);
> args = make_tree_vector_from_ctor (init);
> + args = resolve_args (args, complain);
> + if (args == NULL)
> + return error_mark_node;
> }
> }
>
> diff --git a/gcc/testsuite/g++.dg/cpp1z/class-deduction112.C b/gcc/testsuite/g++.dg/cpp1z/class-deduction112.C
> new file mode 100644
> index 00000000000..8da5868ff98
> --- /dev/null
> +++ b/gcc/testsuite/g++.dg/cpp1z/class-deduction112.C
> @@ -0,0 +1,14 @@
> +// PR c++/106366
> +// { dg-do compile { target c++17 } }
> +
> +#include <initializer_list>
> +
> +template<class T>
> +struct A { A(...); };
> +
> +template<typename T>
> +A(std::initializer_list<T>) -> A<T>;
> +
> +A a{1,2,3};
> +using type = decltype(a);
> +using type = A<int>;
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2022-07-22 21:32 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-07-21 12:43 [PATCH] c++: CTAD from initializer list [PR106366] Patrick Palka
2022-07-21 17:06 ` Patrick Palka
2022-07-22 21:32 ` Jason Merrill
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).