Adventures in Programming Language Bikeshedding

Hey! Thanks for reading! Just a reminder that I wrote this some years ago, and may have much more complicated feelings about this topic than I did when I wrote it. Happy to elaborate, feel free to reach out to me! 😄

"Greg" opens a great little bikeshedder on rust-dev, proving that you can wink and "ironically" be aware that you're doing something you might think is uncool in the general sense while still totally doing it (see "ironic racism"). While much less harmful than marginalizing people, he's suggesting in the subject title ("For Great Justice") and his many qualifications ('I am aware that I am jumping into an issue at a point in time that's considered "late in the game".') that he's aware that his suggestions, no matter how worthy or well-intentioned, are coming from a place where they probably can't or won't be implemented, but he still just can't help himself but send them.

I don't mean to be too hard on the guy -- he's cordial, and I don't want to hate on someone who's just loving what they do, and wanting to spread it. I think the Rust devs did a good job in their responses of giving his requests credence while also declining. The thread makes for a cute read if you're not used seeing these kinds of discussions.

Yin Wang decides to try to shed light into why FP and OOP are both just wrong. I'm guessing (hoping?) the certainty in this statement

Over the years of doing programming languages research, I have seen the answer clearly, so I often find it pointless debating with people on these issues.

is more likely to do with poor use of language than the kind of bravado needed to claim "solving" subjective qualities for a domain as large as computing and the languages behind them.

The article disappointed me: in order to suggest that OOP is "wrong," he writes

You may argue that in Python or Scala, functions are also objects. In Python, every object with the method __call__ is a function. Likewise, in Scala, functions are just objects with a method named apply. But after some careful thought, you will see that this is confusing the fundamental with the derived. Functions are fundamental, but objects containing them are derived. __call__ and apply are themselves the “function objects” you are trying to define in the first place. Python and Scala just kidnapped the functions, jailed them into “objects”, labeled them "__call__" and "apply", and then call them "methods". Of course, when you wrap a function into an object, you can use the object like a function, but this doesn’t mean you can say “functions are also objects”.

But semantically, you don't lose a whole lot, right? I mean, yes, syntactically, writing

(map (λ (x) (transform (string-append "prefix->" (str x))) list))

is way more fun than writing something like

Lists.map(new Function<DomainElement, RangeElement>() {
    @Override
    public RangeElement apply(DomainElement t) {
       RangeElement elem = transformObject.apply("prefix->".concat(t.toString()));
       return elem;
    }
}, list);

but they do about the same thing, right? So saying a "fundamental" is "jailed" doesn't really say OOP is inherently flawed: just that you may not like the definitions.

His analysis on FP is similarly lacking to me -- he states circular data structures are "hard" in Haskell, but

circularList = [4,5,6] ++ circularList

works in one line. It's the kind of statement you make when you read a lot of people fight about programming languages, but don't do a whole lot of work on them yourself.

Bikeshedding is a lot of fun, and we do it for a lot more than languages. It's also critically important: everyone needs to go through a phase where they have passion and opinions before they learn things that might temper them a bit. Not to say that strong opinions are themselves a sign of not knowing things -- if you're lucky, you'll keep your passionate opinions, but have a bit more ammunition to back them. Check out Robert Harper's blog, where he can more or less go to war with Haskell calling it exceptionally unsafe and that its monads are really no big deal compared to SML's, but the author has also written amazing books on type theory..

If we're lucky, Greg and Yin here can get over smartasses like me and continue with their work, and maybe end up writing more bikesheds in the future, backed by field-forwarding scholarship :D