I am the person that would be the one most likely to adopt haskell in the company I work in. Now, I don't and I won't even though I used it for several pet projects and I would say have learned a lot of haskell and associated patterns.
It's not that haskell doesn't have merits. However I feel like they are oversold. While you can build programs that lack certain error classes, they are not bug free and at the same time debugging often was a lot harder than in other languages. The laziness is especially problematic in that regard.
Another problem is cultural. Haskell and Scala community have been the most hostile communities I have experienced. And the friendliness of communities is a sure predictor for growth.
Another important problem is the docs. For some reasons they hardly contain examples. You don't get adoption that way. Even if the maintainers feel that the type signatures are documentation enough.
And as a last remark, the combinatoric complexity of compiler flags is a problem and needs to be fixed. But the ecosystem is in a deadlock there that no compromise could be found on what should be in the language and what shouldn't be. And it kind of indicates a deeper problem in the haskell design process.
It's not that haskell doesn't have merits. However I feel like they are oversold. While you can build programs that lack certain error classes, they are not bug free and at the same time debugging often was a lot harder than in other languages. The laziness is especially problematic in that regard.
Another problem is cultural. Haskell and Scala community have been the most hostile communities I have experienced. And the friendliness of communities is a sure predictor for growth.
Another important problem is the docs. For some reasons they hardly contain examples. You don't get adoption that way. Even if the maintainers feel that the type signatures are documentation enough.
And as a last remark, the combinatoric complexity of compiler flags is a problem and needs to be fixed. But the ecosystem is in a deadlock there that no compromise could be found on what should be in the language and what shouldn't be. And it kind of indicates a deeper problem in the haskell design process.