Algebraic Effects: Another mistake carried through to perfection? | Hacker News Reader

Algebraic Effects: Another mistake carried through to perfection?

kjosib.github.io

29 pointstodsacerdoti11 comments

There’s a certain amount of hubris to say, “I don’t know anything about this and you’re making a mistake.” It’s off putting and kills the whole rant.

I’ve heard opinions from smart people with lots of experience who say algebraic effects are not worth the squeeze. I’ve also heard some say that we should all be pushing the boundaries: they are the future.

So the matter doesn’t seem to be decided. Now isn’t the time for maxims.

The desire of the HN community to pull a random person's uninformed opinion about a topic that they, justifiably, wrote for their own interests and amusement and then pontificate about how either stupid or amazing it is will never ceise to confuse me.

Effects on their own are a very active area of research and I would laugh behind a PL researchers back if they claimed it was a solved issue. Between Monads, call-by-push-value, and algebraic effects there is really no clear "how do people actually use this" answer.

But that's not the job of a PL researcher anyway, or a random software engineer for that matter. Sorry to say, the software engineer knows next to nothing about "the right way" to design language features that people want to use or enjoy using. If anything this should be an HCI person with a penchant for PL or vice versa.

>Effects on their own are a very active area of research and I would laugh behind a PL researchers back if they claimed it was a solved issue. Between Monads, call-by-push-value, and algebraic effects there is really no clear "how do people actually use this" answer.

I can actually say that I used algebraic effects in my thesis for the section on semantics of a basic probabilistic programming language. It avoided talking about monads for my committee member who cared and honestly made for an easier implementation.

> Sorry to say, the software engineer knows next to nothing about "the right way" to design language features that people want to use or enjoy using.

Sorry to say that many PL researchers live in their ivory tower and know next to nothing about things that people care about. One could say that it's not their job, their job is to write papers and get tenure. The number of FP enthusiasts versus the number of large, useful systems written in those languages is all the proof you need.

My statement is a vast generalization and is equally incorrect as the original one.

Anyone who uses words like "ivory tower" I know suffers more from jealousy and anti-intellectualiam than anything else. There is a reason Rust is the most loved programming language of modern times and it's not because they ignored the "ivory tower"

I had to stop and re-read this comment chain because I was sure this was satire

> sweet careers are made of this, so who am I to disagree? Compile the world; Java Python C. Everybody’s looking for some bug. Some of them want to maintain you. Some of them want to be maintained.

For those missing the reference:

https://youtu.be/qeMFqkcPYcg?si=at-YtggekbPdv7sN

Research doesn't work like that. I like the idea of separating contract and implementation in algebraic effects. It might pave the way to bring back some sanity to imperative languages and help us write better code, since it's pretty clear that the "real world" doesn't care much about pure functional languages no matter what they bring to the table. Or algebraic effects could be like monads, many like to talk about them while people building real stuff have no clue about it, nor do they care. But we'll never know unless we explore.

The author says the approach they advocate (just using function parameters) is similar to "dependency injection". It looks like in FP/objects-are-a-poor-man's-closures terminology they're talking about Continuation Passing Style (CPS).

To which implementation is the author referring, I wonder?

I can’t say I recognise any of these issues from freer, polysemy, nor bluefin.

Every time I read stuff like this it just makes me laugh, I honestly never know who to listen to in these debates.

Reply on news.ycombinator.com