我正在尝试使用链接到操作块的有界批处理块。 我知道批处理块中的项目进料结束时我想触发完成链。
问题是:如果我的BatchBlock<T>
属于给定的BoundedCapacity
,我将无法在操作块中触发所有项目。
以下是我的问题的一个示例,它应该(在我对TPL数据流的理解......)中打印0到124,但最终打印0到99.
必须有一些我遗漏的东西......也许BoundedCapacity
意味着“当队列数量超过xxx时丢弃项目......”如果是这样的话,我怎样才能实现有保证的最大内存消耗?
using System;
using System.Collections.Generic;
using System.Threading;
using System.Threading.Tasks.Dataflow;
namespace ConsoleApplication
{
class Program
{
static void Main(string[] args)
{
int itemsCount = 125;
List<int> ints = new List<int>(itemsCount);
for (int i = 0; i < itemsCount; i++)
ints.Add(i);
BatchBlock<int> batchBlock = new BatchBlock<int>(50,new GroupingDataflowBlockOptions(){BoundedCapacity = 100});
ActionBlock<int[]> actionBlock = new ActionBlock<int[]>(intsBatch =>
{
Thread.Sleep(1000);
foreach (int i in intsBatch)
Console.WriteLine(i);
});
batchBlock.LinkTo(actionBlock, new DataflowLinkOptions() { PropagateCompletion = true });
// feed the batch block
foreach (int i in ints)
batchBlock.Post(i);
// Don't know how to end the proper way... Meaning it should display 0 to 124 and not 0 to 99
batchBlock.Complete();
batchBlock.TriggerBatch();
actionBlock.Completion.Wait();
}
}
}
答案 0 :(得分:7)
Post
并不总是成功。它会尝试将邮件发布到阻止,但如果达到BoundedCapacity
则会失败并返回false
。
你可以做的是使用SendAsync
代替它返回一个等待的任务。如果块有空间用于您的消息,则它将异步完成。如果它没有,那么该块返回一个任务,当它有空间接受新消息时将完成。您可以等待该任务并限制插入:
async Task MainAsync()
{
var ints = Enumerable.Range(0, 125).ToList();
var batchBlock = new BatchBlock<int>(50, new GroupingDataflowBlockOptions { BoundedCapacity = 100 });
var actionBlock = new ActionBlock<int[]>(intsBatch =>
{
Thread.Sleep(1000);
foreach (var i in intsBatch)
Console.WriteLine(i);
});
batchBlock.LinkTo(actionBlock, new DataflowLinkOptions { PropagateCompletion = true });
foreach (var i in ints)
await batchBlock.SendAsync(i); // wait synchronously for the block to accept.
batchBlock.Complete();
await actionBlock.Completion;
}