[PATCH] PR libstdc++/90388 fix std::hash<unique_ptr<T,D>> bugs

[Jonathan Wakely]

On 11/05/19 00:51 +0300, Ville Voutilainen wrote:
>On Sat, 11 May 2019 at 00:42, Jonathan Wakely <jwakely@redhat.com> wrote:
>>
>> A disabled specialization should not be callable, so move the function
>> call operator into a new base class which correctly implements the
>> disabled hash semantics. For the versioned namespace configuration do
>> not derive from __poison_hash in the enabled case, as the empty base
>> class serves no purpose but potentially increases the object size. For
>> the default configuration that base class must be kept, to preserve
>> layout.
>
>I continue to not be a fan of the versioned namespace ifdeffery in
>this, but I can live with it.

The versioned namespace configuration should be as good as we can make
it, with no limitations due to ABI compatibility. Removing redundant
base classes allows more compact class layouts. With current trunk
this type has size 3, in the versioned namespace after my patch it has
size 2 (and if we patched hash<optional<T>> it would have size 1):

template<typename T>
struct HashPtr
: std::hash<T*>,
  std::hash<std::optional<T*>>,
  std::hash<std::optional<T* const>>,
  std::hash<std::unique_ptr<T>>
{
};