c# - 将大量可观察量聚合成新的可观察量

Question

比方说，我有 1000 个可观察值。现在我想将所有事件聚合到一个新的可观察对象中，一旦所有其他事件都发送了一个事件，它就会触发 OnNext。使用 Rx 做到这一点的最佳方法是什么？

更新:Rx 论坛上的一些很好的反馈，尤其是 Dave Sexton 的反馈。他展示了如何创建采用多个可观察对象的 Zip 扩展方法:http://social.msdn.microsoft.com/Forums/en-US/rx/thread/daaa84db-b560-4eda-871e-e523098db20c/

Answer 1

F# 中有一个 MailboxProcessor...出于相同的目的，我会在 C# 中使用 SynchronizationContext。给我几分钟，我会写一个例子。

旁白:这是我在 F# 中执行类似操作的代码......这将需要更多的努力，但在 C# 中使用 Rx 仍然可行。

open System.Diagnostics

let numWorkers = 20
let asyncDelay = 100

type MessageForMailbox =
   | DataMessage of AsyncReplyChannel<unit>
   | GetSummary of AsyncReplyChannel<unit>

let main =
   let actor =
      MailboxProcessor.Start( fun inbox ->
         let rec loop acc =
            async {
               let! message = inbox.Receive()
               match message with
               | DataMessage replyChannel -> replyChannel.Reply(); return! loop acc
               | GetSummary replyChannel -> replyChannel.Reply(); return! loop acc
            }

         loop 0 // seed for acc
      )

   let codeBlocks = [for i in 1..numWorkers -> 
                        async {
                           do! Async.Sleep asyncDelay
                           return! actor.PostAndAsyncReply DataMessage
                        } ]

   while true do
      printfn "Concurrent started..."
      let sw = new Stopwatch()
      sw.Start()
      codeBlocks |> Async.Parallel |> Async.RunSynchronously |> ignore
      actor.PostAndReply GetSummary
      sw.Stop()
      printfn "Concurrent in %d millisec" sw.ElapsedMilliseconds
      printfn "efficiency: %d%%" (int64 (asyncDelay * 100) / sw.ElapsedMilliseconds)

      printfn "Synchronous started..."
      let sw = new Stopwatch()
      sw.Start()
      for codeBlock in codeBlocks do codeBlock |> Async.RunSynchronously |> ignore
      sw.Stop()
      printfn "Synchronous in %d millisec" sw.ElapsedMilliseconds
      printfn "efficiency: %d%%" (int64 (asyncDelay * numWorkers * 100) / sw.ElapsedMilliseconds)

main