OCaml事件/频道教程?

R u*_*c k 7 events ocaml channel message-passing

我在OCaml.

我希望模拟通信节点,以查看消息在不同通信方案下传播的速度等.

节点可以1.发送和2.接收固定消息.我想明显要做的是将每个节点作为一个单独的线程.

显然,您可以使用事件模块和通道获取线程以相互传递消息,但我找不到任何此示例.有人能指出我正确的方向还是只给我一个简单的相关例子?

非常感谢.

Tho*_*mas 8

是的,您可以使用OCaml 的Event模块.您可以在在线O'Reilly书中找到它的使用示例.


Vic*_*let 6

如果您要尝试模拟,那么您需要对节点进行更多控制,而不仅仅是使用线程允许 - 或者至少没有重大痛苦.

我对该主题的主观方法是创建一个简单的单线程虚拟机,以便完全控制模拟.在OCaml中这样做的最简单方法是使用类似monad的结构(例如在Lwt中完成):

(* A thread is a piece of code that can be executed to perform some
   side-effects and fork zero, one or more threads before returning. 
   Some threads may block when waiting for an event to happen. *)
type thread = < run : thread list ; block : bool >

(* References can be used as communication channels out-of-the box (simply 
   read and write values ot them). To implement a blocking communication 
   pattern, use these two primitives: *)

let write r x next = object (self) 
  method block = !r <> None
  method run   = if self # block then [self]
                 else r := Some x ; [next ()]
end

let read r next = object (self) 
  method block = !r = None
  method run   = match r with 
                  | None -> [self]
                  | Some x -> r := None ; [next x]
end
Run Code Online (Sandbox Code Playgroud)

您可以创建更适合您需求的原语,例如在频道中添加"传输所需的时间"属性.

下一步是定义模拟引擎.

(* The simulation engine can be implemented as a simple queue. It starts 
   with a pre-defined set of threads and returns when no threads are left, 
   or when all threads are blocking. *)
let simulate threads = 
  let q = Queue.create () in 
  let () = List.iter (fun t -> Queue.push t q) threads in 
  let rec loop blocking = 
    if Queue.is_empty q then `AllThreadsTerminated else 
      if Queue.length q = blocking then `AllThreadsBlocked else 
        let thread = Queue.pop q in 
        if thread # block then ( 
          Queue.push thread q ; 
          loop (blocking + 1) 
        ) else ( 
          List.iter (fun t -> Queue.push t q) (thread # run) ; 
          loop 0
        ) 
  in
  loop 0 
Run Code Online (Sandbox Code Playgroud)

同样,您可以调整引擎以跟踪正在执行哪个节点的节点,保持每个节点的优先级,以便模拟一个节点比其他节点更慢或更快,或者随机选择一个线程在每个步骤执行,以及等等.

最后一步是执行模拟.在这里,我将有两个线程来回发送随机数.

let rec thread name input output = 
  write output (Random.int 1024) (fun () -> 
    read input (fun value ->
      Printf.printf "%s : %d" name value ; 
      print_newline () ;
      thread name input output
  ))

let a = ref None and b = ref None 
let _ = simulate [ thread "A -> B" a b ; thread "B -> A" b a ]        
Run Code Online (Sandbox Code Playgroud)


sni*_*im2 5

这听起来像是在考虑John Reppy的Concurrent ML.似乎是OCaml的类似的东西在这里.

@Thomas给出的答案也很有价值,但是如果你想使用这种并发编程的方式,我建议你阅读John Reppy的博士论文,这篇论文极具可读性,并对CML背后的动机以及它的一些实际例子进行了非常明确的处理.使用.如果您对语义不感兴趣,那么如果您跳过该部分,该文档仍然可读.