1. 24
  1. 1

    that was a really engaging and well written post. i can feel the author’s enthusiasm and excitement about the prospect of a bare-metal lisp revival.

    1. 1

      I’m always concerned when people start talking about WASM extensions. Especially WASM-GC. The whole appeal of WASM (to me) is that it’s small and simple and has everything one needs and no more. If you need most help from the host (to do IO, to do GC, to do threading, whatever) the host can expose any calls to you it wants…

      1. 3

        The main appeal of WASM-GC is that WASM environments are usually sitting next to a heap with ECMAScript and DOM objects, which are usually GC’d. WASM-GC offers the hopeful promise that, in the future, we might write WASM programs which are able to natively interface with those objects without requiring special cases in compilers which target WASM.

        1. 1

          I guess if you’re in a browser that’s true. Is there really much needed there, though? The browser host can already see all of the WASM memory and the WASM itself would use a dummy finalizer, yeah?

          1. 3

            In current Wasm (without the GC proposal), the host just sees a blob of memory. It doesn’t have enough information about the data structures used to support GC inside the Wasm environment. The GC proposal adds this kind of structural info so that the host GC can help.