https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115712
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Version|unknown |15.0
CC| |jwakely at redhat dot com
--- Comment #14 from Richard Biener <rguenth at gcc dot gnu.org> ---
While we now take into account that new/delete might clobber global memory
(without -fassume-sane-operator-new being implemented) we still elide
new/delete pairs. I think that is something the C++ standard allows
(even with insane new/delete operator), so I think this is a testsuite bug.
But I'm not really a C++ expert here (I'd close the bug as INVALID).
