| You seem to be aggressively misreading what I said, stuck in your own "world" and instead of asking for clarifying questions, making unfounded assumptions. > I began by noting that TS has literals and set theoretic types Can you please quote me exactly the part of your original comment that implies that literal types are structural and have subtyping? Because that is what I said. If you did imply that, well, excuse me for helping you by clarifying then. > aggressively trying to prove that TS is somehow "better" than Haskell Sorry, what? Refer to the first line of this comment. Just saying that some type system features are extensions or harder to use in Haskell says nothing about the superiority of TS. This is completely your imagination. Here's me criticizing TS 4 months ago: https://news.ycombinator.com/item?id=46501061 > I wasn't the one to bring Zod into a discussion of type systems... Not sure why this is even relevant, but this was a direct reply to: "...language like TS, which is not type safe at runtime, and which will happily ingest an unexpected value, silently coerce it in all sorts of fun and wacky ways" Zod (just a random library) is a direct counter-example to that. There may be better counter-examples. One just needs a proof of existence - it doesn't have to be good. I then had to repeat that I am also talking about static types because of: "...exhaustive runtime schema for your zero-first non-empty integer array example...". TS can both enforce that as a type (which you never presented for Haskell) and as a value. (nominally typed solutions quickly run into ergonomic problems like phantom types not being composable across library boundaries - eg. refinement types, which are even safer by nature, are structural!) In any case, I'd estimate 99% of people using TS don't encounter any type safety issues caused by the design of TS. Haskell has `unsafeCoerce` too, just a bit wordier than in TS. If wanting to talk about real unsoundness, one would mention something like bivariance (see also: linked comment), but even then almost all of that is entirely irrelevant in most practical software engineering. > (a) I don't understand literals and set theoretic types, despite this whole thread being in reply to a post where I give examples of them in TS; and (b) that I care which type system is "better". Huh, what? Again, I'm thinking none of that. Again, you imply you know better what I'm thinking... > Structural typing would be a gaping hole in Haskell's strict type safety I mean, yeah, let me just repeat OCaml and Scala here, both also famously type safe languages... Why make an argument when there's two immediate counterexamples that were already mentioned? > but I do note I seem to have the edge there This pretty much sums up the difference in attitude, I'm not here to score internet points in an argument. I just wanted to comment on why the world is not black and white and even technically flawed languages like TS have something to learn from. |