Haskell is a research language with lazy evaluation. Also with a lot of fun type-stuff. I guess you could make a language which tries to be immediately pragmati

Haskell is a research language with lazy evaluation. Also with a lot of fun type-stuff. I guess you could make a language which tries to be immediately pragmatic. But it is more focused to try to focus on a relatively small area and see how far you can push those concepts. Haskell is just one datapoint. Are PLs moving in one uniform direction, upwards to some kind of "ideal" state? Arguably not until we manage to collapse all the myriad aspects of both programming languages and things they try to solve, and within which constraints (such as those pesky limited digital computers). Is "math" some formalism that is ideally totally divorced from hardware? Then what about formalisms that try to model some kind of computation? What about linear logic and being able to model the disciplined handling of resources such as memory and handles? Is that incompatible with pure functional programming? Not really, since effectful functional programming is a thing, and effects that are encapsulated (such as mutating a local variable inside a pure function) isn't an effect from the view of the caller of the pure function. What about using types to easily[1] declare the layout of memory, such as using SoA over AoS? I'm sure I've seen some paper on that. [1] Or at least more easily and less ad-hoc than what you typically would do in C/++.