http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60809
--- Comment #6 from joseph at codesourcery dot com <joseph at codesourcery dot com> --- I see nothing that says anything about the sequencing of side effects in initialization expressions with respect to the actual initialization itself, or parts thereof - either to require a particular sequencing, or to make it indeterminately sequenced or unsequenced. That may mean it's implicitly unsequenced, but I'd suggest raising this with WG14.
