Keean Schupke

This: ``` f(callback: (B) => ()) where B: Show ``` Doesn't work because you are not specifying the scope for 'B' you could mean: ``` for f(callback: (B) => ())...

> Use a different algorithm that accomplishes the task from a different perspective. This is not always possible. > Because Word does not need the performance boost. If there are...

> I don't understand why that is important. Please explain. - With the `for` in the outer scope the callback is monomorphic with respect to `f` so the callback can...

Yes but I would rather make the programmer declare a datatype for a polymorphic callback function. This keeps the function syntax simpler, and is called first-class-polymorphism.

> I don't think it is often going to be the case that we want to have a function taking a callback, where both said function and the callback function...

> Okay so the research paper you cited is all about achieving type inference. I disagree, it is all about avoiding having to have higher ranked type annotations. In Haskell...

> How is it simpler? You trade off one thing for another. You avoid nested where clauses in the syntax. The grammar will be simpler, (and this will be visible...

> Edit: also nothing stops you from using the method you prefer in your code and achieving type infererence. I am asking why we must force everyone to abandon function...

> and additionally afaics if I have these in a typeclass object, then I will have need a different implementation for each one No a single record or typeclass object...

> Generally wrapping interacts badly, which is why we want a first-class anonymous union instead of massive case-tree logic to destructure Either. OTOH, first-class without wrapping makes inference undecidable. It...