http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43027
--- Comment #4 from Paul H. Hargrove <PHHargrove at lbl dot gov> --- For what it's worth, the sample input is fine with gcc-3.2.2 gcc-3.3.3 gcc-3.4.0 gcc-4.0.0 gcc-4.1.0 gcc-4.2.1 gcc-4.3.2 and broken with gcc-4.4.2 (the original report) gcc-4.5.1 gcc-4.6.1 gcc-4.7.0 gcc-4.8.0 and still broken with gcc-4.9.0 20131031 I am resisting the urge to prepend "[4.4/4.5/4.6/4.7/4.8/4.9 Regression]" to the bug Summary. I'd be content with resolution as INVALID (same rationale as in PR 37267), but sitting UNCONFIRMED for 3.5 years is getting old.