G++ was omitting code for the exception specification on the lambda
because it was confusing it with an synthesized constructor. This patch
fixes the test to match the comment better; not all artificial methods
are [cd]tors.
Tested x86_64-pc-linux-gnu, applying to trunk.
commit 79adecb0b52f58792a2a2f05394f3df13e2b7fcc
Author: Jason Merrill <ja...@redhat.com>
Date: Sun May 22 16:29:27 2011 -0400
PR c++/47263
* decl.c (use_eh_spec_block): Do use an EH spec block for a
lambda op().
diff --git a/gcc/cp/decl.c b/gcc/cp/decl.c
index 733157a..234daaf 100644
--- a/gcc/cp/decl.c
+++ b/gcc/cp/decl.c
@@ -12759,7 +12759,7 @@ use_eh_spec_block (tree fn)
not creating the EH_SPEC_BLOCK we save a little memory,
and we avoid spurious warnings about unreachable
code. */
- && !DECL_ARTIFICIAL (fn));
+ && !DECL_DEFAULTED_FN (fn));
}
/* Store the parameter declarations into the current function declaration.
diff --git a/gcc/testsuite/g++.dg/cpp0x/lambda/lambda-eh2.C b/gcc/testsuite/g++.dg/cpp0x/lambda/lambda-eh2.C
new file mode 100644
index 0000000..1490a25
--- /dev/null
+++ b/gcc/testsuite/g++.dg/cpp0x/lambda/lambda-eh2.C
@@ -0,0 +1,16 @@
+// PR c++/47263
+// { dg-options -std=c++0x }
+// { dg-do run }
+
+#include <exception>
+
+int main( void )
+{
+ std::set_unexpected( []{ throw 0; } );
+ try
+ {
+ []() throw( int ) { throw nullptr; }();
+ }
+ catch( int )
+ { }
+}
