Hacker News Reader

Comment by BillyTheKing | Hacker News Reader

I agree, I don't really understand everyone's issue with err != nil.. it's explicit, and linters catch uncaught errors. Yes the ? operator in Rust is neat, but you end up with a similar issue of just matching errors throughout your code-base instead of doing err != nil..

View on HN

The problem is that you're forced to have four possible states

1. err != nil, nondefault return value

2. err != nil, default return value

3. err == nil, nondefault return value

4. err == nil, default return value

when often what you want to express only has two: either you return an error and there's no meaningful output, or there's output and no error. A type system with tuples but no sum types can only express "and", not "or".

this is true, but not a problem. Go's pattern of checking the error on every return means that if an error is returned, that is the return. Allowing routines to return a result as well as an error is occasionally useful.

I mean, I wish Go had sum types, but this really isn’t a problem in practice. Every Go programmer understands from day 0 that you don’t touch the value unless the error is nil or the documentation states otherwise. Sum types would be nice for other things though, and if it gets them eventually it would feel a little silly to continue using product types for error handling (but also it would be silly to have a mix of both :/).

Yeah, also you almost always need to annotate errors anyway (e.g., `anyhow!`), so the ? operator doesn’t really seem to be buying you much and it might even tempt people away from attaching the error context.

Reply on news.ycombinator.com