[Bug libstdc++/77686] [6/7 Regression] wrong code on arm-linux-gnueabi and arm-linux-gnueabihf

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=77686

--- Comment #15 from Richard Biener <rguenth at gcc dot gnu.org> ---
Btw, what I fail to find is sth in the (C) standard that specifies the
semantics of an aggregate assignment of union type.  Thus I interpret it to use
the
effective type of the object (determined by the last access) and thus only
transfer the active object (so a union aggregate copy is _not_ a bytewise
copy of size sizeof (union type)).