[patch] fix libstdc++/56278

[Jonathan Wakely]

On 13 February 2013 10:19, Paolo Carlini wrote:
> On 02/12/2013 09:54 PM, François Dumont wrote:
>>
>>     Of course this is not mandatory but I think _Hash_code_base would be
>> cleaner this way only exposing as public what is required by C++11. It can
>> also wait for 4.9.
>
> I like it. Thanks Francois. If Jon has no objections over the next day or
> so, please go ahead.

No objection, I agree it's suitable for 4.8 as it's refining my
regression fix, but I don't think it really makes any difference.
There is nothing required by C++11 for _Hash_code_base's constructors
because it's an implementation detail, and the only way a user can
tell the difference between a public and protected constructor is
trying to construct _Hash_code_base directly, which is obviously
unsupported.