Hi,
we have been accumulating quite a few bugs under the [concepts]
meta-bug, most of which of course aren't regressions. This one is a low
hanging fruit, an error recovery issue where, after some meaningful
diagnostic, tsubst_constraint doesn't know how to handle an
error_mark_node. I believe that ideally we should do better, we should
be able to issue only the first error - I think a TODO in
diagnose_constraint hints at that too - but the below should do for now
(well, in 6.1.0 we issued *3* errors ;-)
Thanks, Paolo.
////////////////////
/cp
2018-02-15 Paolo Carlini <paolo.carl...@oracle.com>
PR c++/84330
* constraint.cc (tsubst_constraint_info): Handle an error_mark_node
as first argument.
/testsuite
2018-02-15 Paolo Carlini <paolo.carl...@oracle.com>
PR c++/84330
* g++.dg/concepts/pr84330.C: New.
Index: cp/constraint.cc
===================================================================
--- cp/constraint.cc (revision 257682)
+++ cp/constraint.cc (working copy)
@@ -1918,7 +1918,7 @@ tsubst_constraint_info (tree t, tree args,
tree
tsubst_constraint (tree t, tree args, tsubst_flags_t complain, tree in_decl)
{
- if (t == NULL_TREE)
+ if (t == NULL_TREE || t == error_mark_node)
return t;
switch (TREE_CODE (t))
{
Index: testsuite/g++.dg/concepts/pr84330.C
===================================================================
--- testsuite/g++.dg/concepts/pr84330.C (nonexistent)
+++ testsuite/g++.dg/concepts/pr84330.C (working copy)
@@ -0,0 +1,12 @@
+// PR c++/84330
+// { dg-options "-fconcepts" }
+
+struct A
+{
+ template<typename T> requires sizeof(T) >> 0 void foo(T); // { dg-error
"predicate constraint" }
+
+ void bar()
+ {
+ foo(0); // { dg-error "no matching" }
+ }
+};