This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: type_info::hash_code()
> Since changing type_info::hash_code() would effectively be an ABI
> change (it would break the requirement that typeid(T).hash_code() in
> separate objects/libraries returns the same value for the same T) I
> think we should rename the current _Hash_bytes code to
> _Murmur_hash_bytes and use that in type_info::hash_code. Then we can
> make _Hash_bytes simply forward to another function, which would be
> _Murmur_hash_bytes for now, but could use siphash or something else in
> future.
Seems fine to me.
There seem to be a steadily accumulating number of runtime-changing
issues with unordered containers. Is there a place where we are keeping
track of this stuff? Ie, bugzilla?
-benjamin