3 ms·
Ask HN: Are Paul Graham's arguments in defending Lisp still valid today?
- AnimalMuppet 10y agoYes and no. Yes: Homoiconicity is a big deal. It allows (Lisp style) macros, which are pretty unique and pretty powerful. No: The argument of the Blub Paradox was mistaken then, and is mistaken now. The mistake is the idea that all languages can be ranked on a one-dimensional axis labeled "power". And (related mistake) that one should choose the most powerful one. To see why this is mistaken, consider Lisp and Haskell. Lisp advocates looks at Haskell, and knows that they are looking down. (How can you get anything done in Haskell? It doesn't even have macros!) But the Haskell types (pun intended) look at Lisp, and they also know that they are looking down. (How can you get anything done in Lisp? It doesn't even have a decent type system!) The correct answer is probably more like thinking of programming languages as a tree in a multi-dimensional space. The program you're trying to write is a vector in that space. Pick the branch of the tree that extends the farthest in the direction of the program you're trying to write. That is, pick the language that will do the most to simplify the task of writing the program. Sometimes that language is Lisp; sometimes it's not. Sometimes it's Haskell; sometimes it's not.
- zzzcpan 10y agoBlub was about productivity features for building products, i.e. writing software faster, not arbitrary features. A decent type system in terms of productivity is certainly not Haskell, they cannot look down on anyone. So the argument is still valid. But sure, sometimes it could be Haskell, but I doubt it, it's probably always faster to not use it, even if you have to come up with a DSL.
- AnimalMuppet 10y ago> A decent type system in terms of productivity is certainly not Haskell, they cannot look down on anyone. I suspect that the Haskell users would disagree with you. (I am not one, so I cannot say with certainty.)