forked from Alepha/Alepha
Split up the evaluation helpers into specific files.
This commit is contained in:
84
Utility/enroll.h
Normal file
84
Utility/enroll.h
Normal file
@ -0,0 +1,84 @@
|
||||
static_assert( __cplusplus > 2020'99 );
|
||||
|
||||
#pragma once
|
||||
|
||||
#include <Alepha/Alepha.h>
|
||||
|
||||
#include <tuple>
|
||||
#include <utility>
|
||||
|
||||
namespace Alepha::Hydrogen::Utility ::detail:: enroll_m
|
||||
{
|
||||
inline namespace exports
|
||||
{
|
||||
/*!
|
||||
* Mechanism to define initializer blocks.
|
||||
*
|
||||
* C++ does not have initializer blocks, but it does have the ability to define variables that are set to
|
||||
* values before `main` is invoked. It is possible to use constructors or functions to leverage this fact
|
||||
* and make blocks of code that run before main:
|
||||
*
|
||||
* ```
|
||||
* struct MyThing
|
||||
* {
|
||||
* MyThing()
|
||||
* {
|
||||
* std::cout << "Hello World, before main!" << std::endl;
|
||||
* }
|
||||
* } beforeMain;
|
||||
* ```
|
||||
*
|
||||
* The above code works such that, as a side-effect of the construction of `beforeMain`, the desired code is
|
||||
* run. However, it is rather cumbersome. There are ways of simplifying this:
|
||||
*
|
||||
* ```
|
||||
* int beforeMainFunction() { std::cout << "Hello World, before main!" << std::endl; return 42; }
|
||||
* const int beforeMain= beforeMainFunction();
|
||||
* ```
|
||||
*
|
||||
* While this is a bit less confusing, there's still a need to define a function and return a dummy
|
||||
* variable. Lambdas with immediate invocation syntax (see `evaluate`) can make this nicer still:
|
||||
*
|
||||
* ```
|
||||
* const int beforeMain= evaluate <=[] { std::cout << "Hello World, before main!" << std::endl; return 0; };
|
||||
* ```
|
||||
*
|
||||
* That is a bit better, but that pesky dummy value is still there. The fact that this is an int is going
|
||||
* to be a potential source of confusion. Instead, we want to keep the reader focused on the fact that code
|
||||
* is being run, and disguise the fact that there's a variable involved.
|
||||
*
|
||||
* `enroll` creates a hook by which the above techniques can be made more clear that it one is running
|
||||
* pre-main code.
|
||||
*
|
||||
* ```
|
||||
* auto myInitBlock= enroll <=[]
|
||||
* {
|
||||
* std::cout << "Hello World, before main!" << std::endl;
|
||||
* };
|
||||
* ```
|
||||
*
|
||||
* Because `enroll <=` precedes the lambda definition it is quite clear that something else is going on
|
||||
* here. The `<=` in this case should be thought of as a `fat left arrow` -- the lambda is being given
|
||||
* to `enroll` to be used as an initializer block.
|
||||
*/
|
||||
inline struct enroll_t {} enroll;
|
||||
}
|
||||
|
||||
template< typename Function >
|
||||
struct registration
|
||||
{
|
||||
explicit registration( Function f ) { f(); }
|
||||
};
|
||||
|
||||
template< typename Function >
|
||||
constexpr auto
|
||||
operator <=( enroll_t, Function &&func ) noexcept
|
||||
{
|
||||
return registration{ std::forward< Function >( func ) };
|
||||
}
|
||||
}
|
||||
|
||||
namespace Alepha::Hydrogen::Utility::inline exports::inline enroll_m
|
||||
{
|
||||
using namespace detail::enroll_m::exports;
|
||||
}
|
||||
Reference in New Issue
Block a user