Roslynator VS silverback

There's also this project https://github.com/JosefPihrt/Roslynator that can add even more analyzers.

It does. Nearly every style guide I've seen for any language that allows this prefers this style. I personally prefer it when such structures are needed, but would suggest following it just for consistency and the practical benefits.

Roslynator.Analyzers

The reason I say 'similar' above is that though you may treat it as production code, there are some things in the nature of writing tests that may require different coding standard to be enforced. A good example of this is the RCS1046 analyzer, enforcing names of asynchronous methods to be suffixed with 'Async'; in this case it would be acceptable to override this rule (with an .editorconfig for example) so that you're method name continues to describe the behaviour of the application, and we don't have tests named like public async Task My_descriptive_method_name_async().

You can write Roslyn analyzer but you can also simply use this set of analyzers: https://github.com/JosefPihrt/Roslynator. It has analyzers for both async method not ending with Async and non-async methods ending with Async.

My suggestion is to look into installing Roslynator or Sonarlint. Both are free and work with either Visual Studio or VS Code. With them you can right click on your solution and Run Analysis and it will give you a list of "code smells" and automatic fixes for many of them. They are great tools for any dev.

Roslynator while not exactly as good as full R# does a pretty good job for the analysis and refactoring hints imo. Paired with Intellicode + NCrunch (paid) so far has been a good enough experience for us.

I prefer the second and roslynator would give RCS1124: Inline local variable1 for the first one.

Roslynator.Analyzers

MassTransit, SlimMessageBus, and SilverBack abstract interactions with message queueing systems. Some of what you are trying to do (I didn't read it deep enough to know exactly what you're using the queue for) would be done as a queue configuration instead (e.g. ordering, de-dupe, etc.).

You can look here to have an idea of the tools and the pipeline (I also use azure devops): https://github.com/BEagle1984/silverback

I tend to declare the parameter as IReadOnlyCollection because I can safely iterate without extra allocations. I know you could just do a ToList by default but it would mean allocating a few arrays in the process, in worst case. You could be smarter and check if the IEnumerable can be casted to IReadOnlyCollection and perform a ToList only when needed, this would be the most flexible implementation. I sometimes create an extension method for this: https://github.com/BEagle1984/silverback/blob/master/src/Silverback.Core/Util/EnumerableAsCollectionExtensions.cs.

I strongly believe in CQRS and all my projects implement it (using a mediator etc. - I even built a framework myself for those stuff https://silverback-messaging.net).

Yes we do use Kafka in the company I work for. I built a library/framework on top of Confluent.Kafka for it. Check it out: https://silverback-messaging.net/ or https://github.com/BEagle1984/silverback.

Silverback! https://github.com/BEagle1984/silverback / https://silverback-messaging.net

😁 Nope, not really. The code displayed is from my “hobby” project: https://github.com/BEagle1984/silverback.

By the way, if you are looking for a good/great library to interact with the brokers -> https://github.com/BEagle1984/silverback