typescript-runtime-type-benchmarks VS Wren

Has been heavily optimized, both in terms of its types and runtime performance. Even including the static parser, many types are about an order of magnitude more efficient than equivalent Zod. Early results show it as marginally faster than any validator currently published to typescript-runtime-type-benchmarks, not including more complex cases where (2) would give ArkType a much more significant advantage.

Zod is a bit of an underdog but it is not fast, AJV which is slightly more common can validate and generate types too but requires using JSON syntax, TypeBox offers familiar syntax to Zod while still being JSON syntax in the background.

First, they're not fast (runtime type checking benchmarks).

Aha, so you're using a library in Java for this. You know about libraries in TS for this, there are plenty of them btw, but you don't use them because it's so easy. Express has `any` type for `req.body` because authors don't care about this either and it's so easy. And TypeScript is the one to blame in that you prefer to work with `any` type for incoming data rather than validating it.

It is so much faster than Zod that Zod basically doesn't show, https://moltar.github.io/typescript-runtime-type-benchmarks/ and according to bundlejs, https://bundlejs.com/?q=zod%2Czod%2C%40sinclair%2Ftypebox&treeshake=%5B*%5D%2C%5B%7B+default+%7D%5D%2C%5B*%5D&config=%7B%22analysis%22%3Atrue%7D, it is even smaller. I genuinely have no clue why Zod is this popular in 2023.

TypeBox is a bit different to other libraries in this space where it's mostly intended to be used with a auxiliary JSON Schema validator. Although it provides a built in JSON Schema compiler (which is currently the fastest (not-AOT) runtime validator available for JavaScript today), it's equally intended to be used with validators like Ajv (or any other standards compliant validator)

I do plan to add some direct comparisons to https://github.com/moltar/typescript-runtime-type-benchmarks as well but haven't had a chance yet.

I think it's more of a case of the extremely low performance bar that's been set by the status quo (for even the simplest of validation structures). There's been a lot of focus on the TS type inference, and less on the runtime performance (which actually matters more as it does reduce operational costs). It probably wouldn't be such an issue if the performance was reasonable, but I mean here's the full breakdown https://moltar.github.io/typescript-runtime-type-benchmarks/.

I found a benchmark for runtime performance, but I haven't found any for intellisense/editor performance.

For a moment I thought it was about wren programming language... [1]

[1] https://wren.io/

For "W" you could use Wren.

Bob Nystrom also has a blog, and his articles are really well written (see his post on Pratt parsers / garbage collectors). I'd also recommend going through the source code for Wren, it shares a lot of code with Lox. Despite the deceptive simplicity of the implementation, it (like Lox) is incredibly fast - it's a great way to learn how to build production grade compilers in general.

Only kind of unrelated ... Every time I see the Bevy logo I'm reminded of Wren language https://wren.io/

If you want a pure C99 (sadly not C89 like Lua) immensely fast embeddable language pure interpreter, wren is a great language with excellent features like overload by arity. There is a huge maturity gap between the languages tho.

Check out the source to Wren: https://wren.io. It’s from the author of Crafting Interpreters and builds directly on what’s discussed in the book (essentially a more complete Lox) and adds several additional types, including an array.

The best C code I have ever read is that of wren.