Yoshifumi Kawai
Yoshifumi Kawai
Single request, PublishAsync each time, would greatly affect the benchmark because no automatic pipelining would be used at all. For practical purposes, I believe that pipelining will work depending on...
It seems easy to change the behavior to make them wait if overflow by using BoundedChannel and add PostPublishAsync method, I'll think about that.
I've started to support command limit and request-timeout and external cancellation. #10
blogged about cancellation and performance https://neuecc.medium.com/patterns-practices-for-efficiently-handling-c-async-await-cancel-processing-and-timeouts-b419ce5f69a4
I implemented it for network(especially redis) at first. Therefore, it was only natural that a KEY was needed. I feel that keyless is difficult as a common interface.
Could you provide benchmark code?
ありがとうございます、そのとおりですね、このままではあまりに不便というのはもっともです……！ いただいた案も含め、ちょっと考えてみたいです。 --- You're right, it's too inconvenient as it is. I'll consider about it.
fixed at 1.7.0.
I will look into it on a priority basis.
important...!