http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56217
--- Comment #4 from Jim Radford <radford at blackbean dot org> 2013-02-06 20:46:42 UTC --- (In reply to comment #3) > Should be fixed on the trunk Thanks! > Note your testcase is invalid, without > the taskwait I've added, it relies on NRV and even with NRV, by the time the > task can run the temporary result of f() can be destructed already. I had the taskwait in my original. Still, thanks for the explanation; it adds to my nascent mental model of openmp. Maybe one day I'll get something to compile so I can try it out, but first I'm off to compile gcc!
