From: Richard Peters (R.A.Peters_at_[hidden])
Date: 2001-12-06 10:09:42
Are the boost::bind() versions acceptable replacement for those classes? I
think they nicely fit in the functional library. Their use could be much
more intuitive than the bind() alternatives.
Richard Peters
From: "Peter Dimov" <pdimov_at_[hidden]>
> From: "Richard Peters" <R.A.Peters_at_[hidden]>
> > What about a more general case: a constant generator, and a
unary_function
> > that ignores the first value but returns the generator instead? Combined
> > they provide the functionality of your Constant, and they are also
usable
> as
> > stand-alone functors. And maybe the generator class that behaves similar
> to
> > unary_function/binary_function (doesn't it already exist?)
> >
> > template<class Result>
> > struct generator {
> > typedef Result result_type;
> > };
> >
> > template<class Arg, class Result, class Generator>
> > struct unary_ignore: public unary_function<Arg, Result> {
> > unary_ignore() {}
> > unary_ignore(Generator thegenerator):
> > generator(thegenerator) {}
> > Result operator()(typename call_traits<Arg>::param_type)
> > {
> > return Generator();
> > };
> > private:
> > Generator generator;
> > };
>
> boost::bind(thegenerator);
>
> > template<class Result>
> > struct constant: public generator<Result> {
> > constant(Result thevalue):
> > value(thevalue) {}
> > Result operator()()
> > {
> > return value;
> > }
> > private:
> > const Result value;
> > };
>
> boost::bind(identity<Result>(), value);
>
> > and the same for the binary_function, you could ignore one of the two
> > parameters:
>
> Ditto.
>
> --
> Peter Dimov
> Multi Media Ltd.