i love the fact that mads (who, lets not forget is a C# language designer) said "a bunch of people claim they understand Monads but in reality most don't and if you do understand it, its not for long" Then he says he has understood it many times and then the next day was like "what was it about again?" - How awesome and humbling to hear.
I'm pretty sure that Mads knows exactly what they are and this was slightly tongue in cheek :-) But yes, there's a lot of confusion about them (and the misunderstanding that you need to know what they are to "do" FP).
Coming from a functional background, this talk did a lot to improve my understanding of c sharp. Related it back to what I know and pointed out a bunch of equivalences that I had missed. I don't think I'll ever be a very good c sharp coder but at least it showed me how to read the language. I can relate to having "understood monads many times" lol. Really enjoyed the presentation :)
Compiler language features vs. runtime language features would be a nice thing to have control over. You either set your lang version above what your runtime supports and you have to keep track of runtime features and not use them or you keep it current and can't use any new syntax sugar. Unless this already exists and I've never been able to find it 🙃
I wish the csharp language team would tackle the null clusterfuck with more effort. The nullable reference types are a neat compiler feature but produces just warnings and enforces nothing. Maybe give us code contracts for method signatures. Like "this param cannot be null" or "this integer must be between 0 and 100" or "will throw this exception when this condition is met". XML documentation is nice (if existing at all) but having a compiler that enforces and safeguards APIs would be way better. Remember ".NET Code Contracts" from Microsoft Research? Pepperidge Farm remembers.
You are very correct. I tried it a bit. Noticed that it doesn't enforce anything. Then I completely disabled it. It would be very very nice if it was a very strict system.
@dasunguy @XtroTheArctic "just warnings" by default. Toss something like dotnet_diagnostic+CS8629+severity = error into an editorconfig and voila, enforcement. (Replacing '+' with period). CS8618 and CS8602 are other relevant ids you may see. Fine-grained control (silent/suggestion/warning/error) is possible, no need for a blanket "all warnings as errors" Edit: Or via VS UI, click lightbulb => Suppress or Configure Issues => Configure Severity ... writes it to editorconfig
This is nice but I just wish F# was more popular and accepted. I am doing debugging on an objected oriented codebase I was not involved with building that has mutable global objects injected all over the place and it's just a nightmare.
When I close my eyes, I see parameterless void methods doing internal mutation. The spaghetti becomes so unbearable, that it makes you wish things were declared explicitly mutable or immutable. But then there are functional patterns to save us, along with Rust for those who like it.
@@zhh174 They are trying to get some other features out be proceed with DU - as DU would be built on top of them. Hopefully it would be done sooner rather than later
I am lookin at Zoran Horvat's course on Pluralsight in C# 10 and it seems like records (and record structs) in C# 9 are another important building block in making C# a bit easier to use with functional programming style.
@@kooraiber I believe he was the original leader of the team that developed C#. Before that, he worked for Borland and was the leader of the team that developed Delphi.
The best part about csharp becoming more functional is that people's minds will be more open to Fsharp lol. I made the switch this last year and I am so unbelievably happy it's unreal.
