[RFC] Shall we provide 3 overloads for istream::ignore?

[Paolo Carlini]

Hi,

according to 27.6.1.3, p24, the signature of istream::ignore is:

  ignore(streamsize n = 1, int_type delim = traits::eof())     [1]

As often happens when defaults are involved, in this case too, providing,
additionally to

  ignore(streamsize n, int_type delim)                         [2]      

the overloads

  ignore(streamsize n)                                         [3]

and
   
  ignore()                                                     [4]

should lead to much improved performance: [4] would be obviously trivial,
no loop involved; [3] may take into account that delim is traits::eof(),
and then no delim at all would be searched in the buffer.

Therefore, may question: is there anything wrong with providing those
overloads? What about the ABI? Is it completely safe providing [2], [3] and
[4] instead of simply [1]?

Thanks,
Paolo.