Hi,
in this ICE on valid, 4.8/4.9 Regression, the problem is that, for a
lambda in an NSDMI context, the early return of lambda_expr_this_capture:
/* In unevaluated context this isn't an odr-use, so just return the
nearest 'this'. */
if (cp_unevaluated_operand)
return lookup_name (this_identifier);
fails to find a this_identifier. This is expected, in a NSDMI context,
as also clarified from the comment a few lines below. Thus I'm handling
the issue the same way, that is by way of scope_chain->x_current_class_ptr.
Tested x86_64-linux.
Thanks!
Paolo.
////////////////////////////
/cp
2013-10-17 Paolo Carlini <paolo.carl...@oracle.com>
PR c++/58596
* lambda.c (lambda_expr_this_capture): Handle NSDMIs in the
cp_unevaluated_operand case.
/testsuite
2013-10-17 Paolo Carlini <paolo.carl...@oracle.com>
PR c++/58596
* g++.dg/cpp0x/lambda/lambda-nsdmi5.C: New
Index: cp/lambda.c
===================================================================
--- cp/lambda.c (revision 203744)
+++ cp/lambda.c (working copy)
@@ -628,7 +628,14 @@ lambda_expr_this_capture (tree lambda)
/* In unevaluated context this isn't an odr-use, so just return the
nearest 'this'. */
if (cp_unevaluated_operand)
- return lookup_name (this_identifier);
+ {
+ /* In an NSDMI the fake 'this' pointer that we're using for
+ parsing is in scope_chain. */
+ if (LAMBDA_EXPR_EXTRA_SCOPE (lambda)
+ && TREE_CODE (LAMBDA_EXPR_EXTRA_SCOPE (lambda)) == FIELD_DECL)
+ return scope_chain->x_current_class_ptr;
+ return lookup_name (this_identifier);
+ }
/* Try to default capture 'this' if we can. */
if (!this_capture
Index: testsuite/g++.dg/cpp0x/lambda/lambda-nsdmi5.C
===================================================================
--- testsuite/g++.dg/cpp0x/lambda/lambda-nsdmi5.C (revision 0)
+++ testsuite/g++.dg/cpp0x/lambda/lambda-nsdmi5.C (working copy)
@@ -0,0 +1,7 @@
+// PR c++/58596
+// { dg-do compile { target c++11 } }
+
+struct A
+{
+ int i = [] { return decltype(i)(); }();
+};