Lesley LaiNov 24

New blog post: "Fifty Shades of OOP." Put in a surprising amount of research for this one!

https://lesleylai.info/en/fifty_shades_of_oop/

Fifty Shades of OOP | Lesley Lai

This post talks about the many different aspects under the umbrella term OOP

Show threadKartik AgaramJan 1

@lesley @aartaka One thought I've had for a while is that programming languages and the rhetoric around it focus too much on tangible things like member variables and methods.

David Parnas was the first to point out [1] that each piece of a system -- whether you call it a function or method or object or class -- should exist to "protect" a single coherent invariant. That's the thing we need to ensure we preserve in the course of extension, not any particular member or method.

[1] David Parnas, "On the Criteria To Be Used in Decomposing Systems into Modules," 1972

Show threadLesley Lai

@akkartik @aartaka Sounds like the syntax vs semantics debate. This reminds me of the "Wadler's law." That said, while we probably spend too much time bikeshedding syntax, I do think it matters because that's how _human_ interface with a programming language. Somewhat related:

https://slim.computer/concrete-syntax/

Concrete syntax matters, actually

Jan 2 at 5:47amWeb
Show threadKartik AgaramJan 2

@lesley Hmm, I think this is more about pragmatics. I want the syntax of a programming language to guide you to think more clearly about the problem. Should a member variable be public or private? It's irrelevant. What matters is whether an invariant is violated at some point in a program. Languages should better emphasize invariants.

Hmm, I wonder what Eiffel did? No, https://en.wikipedia.org/wiki/Eiffel_(programming_language)#Scoping also seems weird. So I'm suggesting some speculative new research direction.

Eiffel (programming language) - Wikipedia

Show threadLesley LaiJan 2

@akkartik Ahh. I misunderstood what you said.

D also supports an `invariant` keyword. I wonder whether this has a performance implication (do they assert the invariant all the time?) It is kind of interesting, but I don't see many languages following Eiffel's footsteps.

Show threadKonrad HinsenJan 4

@akkartik That's something that has been on my wish list as well for quite a while.

I'd like a language to let me express whatever I know or want to be true of a software system, irrespectively of whether some tool (compiler, type checker, ...) can make use of it.

@lesley