Comment by _5yoy | Hacker News Reader

Comment by _5yoy | Hacker News Reader

Parent Full thread

> You are making a very common complaint, that comes from ignoring the problems of the tools you know while making a fuzz over anything you find on the new one. It's almost certain that whatever languages you are used to, they have a much larger pile of gotchas than Haskell (because Haskell has an atypical small number of them), but that doesn't stop them from being useful.

"Yeah well your language isn't good either" isn't a very good argument for why a language is _better_. If the idea is "Haskell works better than other languages as long as you happen to know how to make it so", isn't that just every language that exists? Due to Haskell's branding, I was under the impression that the issues the above user is describing should be impossible. If Haskell doesn't even make good on those guarantees, but instead shifts that guarantee as "Things you should've known not to do" onto the user, how is that different from any other language?

edit And since this thread has a lot of tension and vague hostility, I will add, I am not attacking you or the language you enjoy. I legitimately would like to better understand.

Hum... I do think "Haskell has an unusually low amount of gotchas" isn't a subjective argument that applies to any language. Also, no, yours "Haskell works better than other languages as long as you happen to know how to make it so" can't apply to every language, by construction you can only put one language on it¹, unless you assume that Haskell is the only language the developer knows well, what would make for a really lame argument.

But "you won't be able to use it well if you don't learn how to use it" does indeed apply to all languages. Yet, it's more relevant for some, and Haskell is one of those more affected by it.

It is a fact that Haskell is not beginner friendly. Your first program on it will suck, whatever proficiency you have with other languages (there are some people that argue that it is easier if you know less beforehand). It takes some learning before you are able to get the better safety, high productivity and easy collaboration people talk so much about (and even then, you won't get those for every problem, the language has limitations too).

1 - The statement is false anyway. There's no such language one can single out there without more context.

Code written in Haskell is only safe if the authors of such code take advantage of the features that the language provides to write safe code. Haskell makes it ergonomic to write safe code, not mandatory. It turns out that the authors of Haskell's Prelude 30 years ago didn't get things quite right. We're still dealing with that today and it takes time and effort to fix things.

I understand your point. My question, though, is: for what languages is this statement not true:

" < language > is only safe if the authors of such code take advantage of the features that the language provides to write safe code "

It's not a binary classification of "Haskell safe, everything else unsafe". It's a gradation. Haskell provides more such features and more ergonomically (goes the claim, and it's one I happen to agree with).

Well, while python still has 2 or 3 gotchas, I think it can be called safe (now the question for me arises what do we actually mean by safe).

Reply on news.ycombinator.com