https://gcc.gnu.org/bugzilla/show_bug.cgi?id=85544
--- Comment #7 from Thomas Koenig <tkoenig at gcc dot gnu.org> --- The problem with the patch in comment #6 is that it pessimizes code like b = (-1)**a So, some more thinking required...
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=85544
--- Comment #7 from Thomas Koenig <tkoenig at gcc dot gnu.org> --- The problem with the patch in comment #6 is that it pessimizes code like b = (-1)**a So, some more thinking required...