Tony Morris

Well, my team uses nix, but I would also never burden a student with that either. There are better solutions to stack in every context, but it's not the same solution in each of those contexts.

Context: Haskell learner wants to write a small project that uses a few Haskell libraries from hackage. "Nix" is too much of a burden to learn simultaneously. "cabal install" fails cryptically. What should the learner do?

So, now we are imposing anti-learning onto arbitrary students? What if the student is not anti-learning? That's not a context. Cabal install does not fail any more cryptically than other shitty build tools.

cabal install fails cryptically far more often than stack does, and that's the point. There are a lot of people who want to give haskell a shot. Their budget (in time, energy) for learning is understandably limited, they'd rather spend it on learning haskell than nix.

Suppose this were true (it's not, but suppose it was). Why does every critique of stack turn into, "but it's not cabal"? Here is some news for you: I HATE CABAL MORE THAN YOU DO. Can we please stop talking about cabal? Please, fucking please, goddamn motherfucking please?

talking about whether stack is good or not necessarily compares it to its alternatives. Its goal of installing haskell software is not contested, so the question is whether it's better than its alternatives at this purpose or not. Naturally this leads to discussing alternatives

The alternative to stack is not http://stack.Talk  about that.

not stack fails to install any existing haskell libraries, unless it happens to be of the "nix" or "cabal" variety. stack is far superior to both in important aspects (not all, of course) pedagogy isn't everything.