Re: [boost] concepts and template template parameters (was: [Proto] Review)

From: Giovanni Piero Deretta (gpderetta_at_[hidden])
Date: 2008-03-31 10:22:43

On Mon, Mar 31, 2008 at 2:25 PM, Douglas Gregor <dgregor_at_[hidden]> wrote:
>
>  On Mon, 2008-03-31 at 00:09 -0700, Eric Niebler wrote:
>  > [...] The question is whether this should compile in C++0x:
>  >
>  > // Takes an unconstrained template:
>  > template<template<typename T> class S>
>  > struct foo
>  > {};
>  >
>  > // A constrained template
>  > template<Regular Value>
>  > struct vector
>  > {};
>  >
>  > foo<vector> f; // OK? Prolly not.
>
>  Yes, this is well-formed C++0x. We (have to) let unconstrained template template parameters bind to constrained
> templates.

This is good news for metaprogramming

but in general it might reopen the problem of deferred typechecking of
template parameters.
Are constrained template template parameters legal?
i.e.:

  // Takes a constrained template:
  template<template<Value> class S> // is this syntax even legal?
  struct foo {};

  // A constrained template
  template<Regular Value>
  struct vector
 {};

  foo<vector> f; // OK?

[ If it is not legal, I guess you could fake it with concept achetypes
and some kind of rebind trick.]

Not useful for metaprogramming, but for more mundane things, you might
not want to lose the benefits of
concept typechecking.

-- 
gpd