Devvy / Kepler @68kHeart@tech.lgbt

I would be like "oh, this is probably some .NET stuff they have to work around", but if the solution is legit just "make it a function that takes unit", that's something the compiler could just abstract away for you. And yet,

Oh. That's why so many things in F# just take unit as an argument instead of just being a plain value: F# does not allow values to be generic, only functions. Boy, this single-pass compiler stuff is the gift that keeps on giving. /s

Trying to do a project in F# because I'm likely going to need .NET fuckery down the road. It…sure is a language! I love that they're like, "yeah, we wanted an OCaml-style language because of the safety it provides, but also you shouldn't try to capture all failures and represent them as types and instead throw exceptions", like ??? doesn't that defeat the purpose of choosing an ML-style language to base your work off of??? Even Haskell devs are like "yeah, exceptions were kind of a big mistake"

There's an experimental try-block syntax that does address the issue of capturing errors from the question mark operator. But that's like, great, now we're back to wrapping everything in try blocks again because error handling is a problem!

You have many useful functional programming concepts, like .map() and .and_then() at your disposal, but then you have to define lambdas for operations because everything is methods — oops!

The question mark operator really drives home how much of an issue this is. But even then, it's still very hard to use in a way that lets you handle and structure errors nicely.

I'm trying Rust again and I think I realize why I keep hitting a wall with it:

1. It uses a lot of concepts from category theory and already established functional programming languages, and then…just doesn't use them in a way that's helpful; and this is made worse by

2. Lots of macro magic, either by crates or by the compiler, that is not always intuitive and opaque in how it works.

still better than Adobe Illustrator tho tbh