I Don't Use Exceptions in C++ Anymore
I Don't Use Exceptions in C++ Anymore
Using exceptions in C++ desktop and server applications overall made sense to me. As I expanded my usage of C++ into other domains, specifically embedded domains, I began to experience more compelling reasons not to use exceptions first-hand...
From lobste.rs
You're viewing a single thread.
It's a shame that sum type support is still so lacking in C++. Proper
Resulttypes (ala Haskell or Rust) are generally much nicer to deal with, especially in embedded contexts.As is, there's only
std::expected(which can and will blow up in your face if you forget to checkhas_value) andstd::variant, which I have heard nothing but complaints about.It’s a shame that sum type support is still so lacking in C++. Proper Result types (ala Haskell or Rust) are generally much nicer to deal with, especially in embedded contexts.
I don't think this is a lack of support in C++. There are already a few C++ libraries that implement Either and Result monads. It would be nice if those were supported in the C++ stand library, but that does not stop anyone from adopting them.
I would consider language support essential for "good" sum types. AFAIK, stuff like exhaustive pattern matching can't be accomplished by a library. Perhaps you could do some cursed stuff with compiler plugins, however.
(There was a library that implemented non-exhaustive pattern matching that eventually morphed into an ISO C++ proposal... so we won't see it until 2030 at the earliest /hj)
At a library level, couldn't you have an opaque sum type where the only thing you can do with it is call a
matchmethod that requires a function pointer for each possible variant of the sum type? It'd be pretty cursed to use but at least it wouldn't require compiler plugins.I'd bet five bucks some desperate Haskeller or Rustacean has implemented exactly that. You could also probably use nested functions for GCC C or lambdas in C++ to move everything inline?
I think that pattern matching and sum types are orthogonal to monads, and aren't really relevant when discussing monads as alternatives to exceptions. C++ didn't required any of those to add std::optional or std::variant, and those are already used as result monads.
Supporting Result and Either monads in the standard would be nice, but again this does not stop anyone from adopting one of the many libraries that already provide those.
Well, if you create result types without monads, you get go.
I would say it's completely essential, but you can do with some limited implementation of them.
I guess it depends on what you mean by using monads, but you can have a monadic result type without introducing a concrete monad abstraction that it implements.
You seem to be ignoring the benefits of compiler support as mentioned in the comment above.
compiler support
That remark was on sum types, not monads. You do not need "compiler support" to have Result or neither monads in C++. There are already plenty of libraries that implement those. I use them in some of my projects. No compiler support needed.
As I said, sum types are not required for Return or Either monads. At best, they are convenient.
The original claim wasn't that you can't implement monads in C++, it was that compiler support is needed for "good" sum types. Unless I'm misreading, you brought monads into it. And they're not totally orthogonal: sum types are a very good way to implement monads.
Unless I’m misreading, you brought monads into it.
You're misreading it. What do you think a '
Result' type is?Well, in Rust, it's a sum-type, with functions that also let you use it like a monad instead of using explicit pattern matching.
Well, in Rust, it’s a sum-type
The discussion is on to use monads in C++, and not on why is C++ different than Rust.
I repeat: you do not need sum types to implement a Result monad in C++.
The discussion was about sum types. The top-level comment, the one to which you originally responded, says:
It's a shame that sum type support is still so lacking in C++. Proper
Resulttypes (ala Haskell or Rust) are generally much nicer to deal with, especially in embedded contexts.