Re: [boost] Yap's formal review is starting now!

Subject: Re: [boost] Yap's formal review is starting now!
From: Peter Dimov (lists_at_[hidden])
Date: 2018-02-16 16:22:21

Steven Watanabe wrote:

> The thing is, in most cases, the tuple type is chosen by Yap, not by the 
> user.  Most of the time, users don't actually need to care what the tuple 
> type is, either.

...

> struct a_transform {
>   template<class T>
>   auto operator()(terminal_tag, T&&t);
>   // ...
> };

Right.

I'd guess that tag matching is a later addition, because it only works for 
simple cases. You can't do

    auto operator()(any_expression_tag, T&&... t);

because there's no way to match a tag, they are just classes, unrelated to 
anything (and do not even appear in the reference.)

One would expect instead of

    struct plus_tag {};

something rather like

    template<expr_kind kind> struct expression_tag {};

    using plus_tag = expression_tag<expr_kind::plus>;

and now f.ex.

template<expr_kind kind, class... T>
auto get_arity::operator()(expression_tag<kind>, T const&... t) const
{
    return std::max( { yap::transform(t, get_arity{})... } );
}