On Mon, 29 Aug 2016, Marek Polacek wrote: > On Fri, Aug 26, 2016 at 02:51:38PM -0700, Paul Eggert wrote: > > > Well, the patch could use __SIZEOF_FLOAT128__ just as well as __i386__ > > > (the effect would be an extra union member > > > > s/union/struct/. Though I've always wondered why it is a struct and not a > > union. Maybe change it to union while we're doing an ABI change anyway? > > Yeah, me too. The initial implementation is here > <https://gcc.gnu.org/ml/gcc-patches/2011-11/msg00841.html> but I don't see any > comments wrt max_align_t being a struct or a union there.
I'm not aware of a specific reason for struct versus union (naming via a typedef without a struct tag, so that no identifiers from the tag namespace are used for C and so that the name for linkage purposes in C++ is max_align_t, is deliberate, however). While the chance of any code's ABI being affected by the size of the type should be small, the minimum change is certainly the one that uses __i386__ and so doesn't affect the type at all except in the case where it's necessary to do so. -- Joseph S. Myers jos...@codesourcery.com
