SO CSharp

This channel was never meant to be just about stolen memes and jokes. One of the goals was to share some random pieces of knowledge (#random_PoK 😁) that I come across. Since I mostly work with .NET, they’ll likely be .NET-related — but we’ll see.

So, here's one of them — a niche but interesting Task.WhenEach method. Added in .NET 9, it allows iterating over tasks as they complete via the async enumerable it returns. Here's a basic example:

List<Task<string>> tasks = [ .. ]; // some tasks

await foreach (var completed in Task.WhenEach(tasks))
{
    Console.WriteLine(completed.Result);
}

I haven't found many production use cases for this yet but one I can imagine is processing a subset of tasks until some condition is met — similar to this SO question, which otherwise would require more verbose code.

I also composed a simple benchmark which showed no significant runtime difference compared to Task.WhenAll. However, the higher completed work items count and lock contentions are worth noting:

| Method       | Mean     | Error    | StdDev   | Completed Work Items | Lock Contentions | Allocated |
|------------- |---------:|---------:|---------:|---------------------:|-----------------:|----------:|
| TaskWhenAll  | 15.58 ms | 0.086 ms | 0.081 ms |              11.0000 |                - |   3.99 KB |
| TaskWhenEach | 15.53 ms | 0.068 ms | 0.061 ms |              16.6250 |           0.0156 |   3.82 KB |

That said, benchmarks like this are synthetic, so always test your actual workload on real hardware. There's also the question of how applicable microbenchmarking is to parallelized code 😊.

#csharp #dotnet

👍6🥰1

61 viewsedited 09:12

SO CSharp

Those are the only meaningful numbers regarding the #php experience 😁

#belittlinglangs

❤5🤣2

78 views12:15

SO CSharp

#refactoring #legacy

😁3💯2❤1

52 views04:07

SO CSharp

#InfoSec