Ruby website redesigned

Yeah, I thought those code samples would run immediately, in which case maybe the loading would be justified (although surely very easy to avoid). Instead, they're links to a different page that has the same code sample and a link to run the code, meaning I need to press twice to see what the code does when it runs, which isn't a lot but is surely at least one (possibly two) clicks more than necessary.

That said, it's cool seeing some of those samples, because they're honestly not really what I expected. For example, I didn't expect the list subtraction to work at a set operation, so seeing that example gives me a feel for what sort of things I can do with Ruby code.

I like how it looks. I don't like to see how badly it is crafted tech-wise - not optimized images by size and deferring, JS for things that work natively in the browser, bloat of tailwind instead of nice clean and modern CSS.

Knowing ruby I can tell that the relaxed approach to the website does not correspond with sophistication in the language itself. If I wouldn't know ruby, that would be a put off for me, thinking that if they don't want to convince me tech-wise by their site, it might be similarly annoying to deep-dive into the language.

The Lighthouse report is telling. It scores 100% for Best practices and SEO, but 54% for Performance. Pages like these used to be caricatures of the modern web, but are now acceptable. DHH's statement doesn't help either.

Ruby has had YJIT for some time and being deployed and used in production, from Github to Shopify.

The current experimental JIT is ZJIT. And the fastest Ruby JIT Runtime is TruffleRuby. ( I wish JRuby gets more love )

> somehow, the Ruby community [...] think it's acceptable to have a standard implementation that does neither AOT nor JIT native code compilation

Ruby have YJIT, which is a production ready JIT compiler that generates native machine code. But it requires enabling via flag "--yjit" rather than running by default.

Why? I think it's primarily to avoid build time dependencies on Rust and prevent unexpected overhead for users. This keeps binary light and avoids forcing Rust installation on users, especially for those who run interpreter only, where YJIT adds no value.

Note that including YJIT also bloat binaries by 5 to 10MB (Rust static lib + code cache structures) for source builds and complicates cross compilation since Rust targets vary by architecture (focus x86-64 and arm64, not all platforms).

Also, Rails 7.1+ enables YJIT by default, so JIT (to native code) in Ruby is being utilized when actually needed.