@kaleidic And yet if you believe that all homelands are equivalent, and all rulers equivalent, maybe we should trade you for a North Korean who disagrees.

@kaleidic There is no reason to believe that the new ones will be criminals, even less to believe that if they become so, they will be worse than the current ones. And plenty of reasons to stop (and if possible punish) previous criminals rather than leaving them in place.

@miehrmantraut OO and FP work great together. Typeclasses FTW!

@TheEduardoRFS @monadivalence As Wadler says "Theorems for free!" At least initially...

@vpatryshev I guess I'm somewhere else. Wondering what kind of categories correspond to my reflective ideas. Did you my thesis? Or can you read it and tell me?

@TheEduardoRFS @monadivalence There are sweet spots for types when they say enough to be useful, but not so much that you're replicating the complexity of the program in a different paradigm (which then costs more like 10x). Yet when correctness is life-and-death, you may afford the semantic replication.

@monadivalence @TheEduardoRFS Type-driven code generation is great PRECISELY because you didn't need to move ALL the semantics into the typesystem, and types remain a very poor approximation of the semantics.

RT @Ngnghm: @monadivalence @TheEduardoRFS When your types become so precise that you can only use the correct variable the correct number o…

@the_octobro I'd love to see someone implement that in Erlang—but since Erlang is locally eager and the construct needs laziness to not explode exponentially, I suppose that means at least one process per object to handle the late binding (or a per-process registry for objects). Yikes.

@the_octobro Unhappily most languages only adopt some low hanging fruit features from the "other" paradigm while frustratingly missing key features and not fitting well with rest of the language. No inheritance in Erlang or in Haskell typeclasses. Limited lambdas in C++, Java or Python.

@chillbert32 Yes, but also the clever breaking aparts are often problem-dependent and not reusable in other circumstances, whereas the clever chippings-away often are. That's why we have reusable (type)class libraries.

If you can reduce your problem, you probably should... though at the same time, it could be introducing notions so abstract (to you) that they boggle the mind, at which point that would slow you down instead of help you. Hence an IQ bias toward FP. Still wrong to dismiss OO.

@vpatryshev I don't understand either, but I'd guess that's a very Kleisli category mental model way of putting it.

@AndrewCMyers @krismicinski One of us did.