I was mistaken to assume that a dependent type is necessarily
incomplete, and indeed there are multiple places in the frontend where
we check a type for both dependency and completeness. So this patch
partially reverts the fix for PR95497, restoring the dependent_type_p
check that guarded the call to is_really_empty_class below.
gcc/cp/ChangeLog:
PR c++/96132
* constexpr.c (potential_constant_expression_1) <case PARM_DECL>:
Restore dependent_type_p check that guarded the call to
is_really_empty_class.
gcc/testsuite/ChangeLog:
PR c++/96132
* g++.dg/template/incomplete12.C: New test.
---
gcc/cp/constexpr.c | 1 +
gcc/testsuite/g++.dg/template/incomplete12.C | 9 +++++++++
2 files changed, 10 insertions(+)
create mode 100644 gcc/testsuite/g++.dg/template/incomplete12.C
diff --git a/gcc/cp/constexpr.c b/gcc/cp/constexpr.c
index ff78ebda2dc..97dcc1b1d10 100644
--- a/gcc/cp/constexpr.c
+++ b/gcc/cp/constexpr.c
@@ -7444,6 +7444,7 @@ potential_constant_expression_1 (tree t, bool want_rval,
bool strict, bool now,
{
tree type = TREE_TYPE (t);
if ((processing_template_decl && !COMPLETE_TYPE_P (type))
+ || dependent_type_p (type)
|| is_really_empty_class (type, /*ignore_vptr*/false))
/* An empty class has no data to read. */
return true;
diff --git a/gcc/testsuite/g++.dg/template/incomplete12.C
b/gcc/testsuite/g++.dg/template/incomplete12.C
new file mode 100644
index 00000000000..335e5356874
--- /dev/null
+++ b/gcc/testsuite/g++.dg/template/incomplete12.C
@@ -0,0 +1,9 @@
+// PR c++/96132
+// { dg-do compile }
+
+template <int> class a;
+
+template <int b> class c {
+ a<b> e;
+ void operator=(c d) { e = d; }
+};
--
2.27.0.343.g4a0fcf9f76
