←back to thread

Web Embeddable Common Lisp

(turtleware.eu)
138 points todsacerdoti | 10 comments | 25 Jun 25 15:33 UTC | HN request time: 1.2s | source | bottom
1. bevr1337 ◴[25 Jun 25 19:09 UTC] No.44380860[source]▶
Why was LISP not one of the first languages able to compile to WASM? I've been curious ever since the awesome WASM project started tracking different projects. They use emojis for progress and LISP has been a hatching egg.
replies(2): >>44381096 #>>44381862 #
2. no_wizard ◴[25 Jun 25 19:37 UTC] No.44381096[source]▶
It sorta has been. As a language design it is already native via the WAT[0] which compiles to web assembly.

[0]: https://webassemblyman.com/wat_webassembly_text_format.html

replies(1): >>44381200 #
3. bevr1337 ◴[25 Jun 25 19:47 UTC] No.44381200[source]▶
Do you know why only sorta? My understanding is that only toy implementations exist right now but nothing production ready. I'm poking around Google and GitHub and not finding any new information.

> Implementing Lisp onto WebAssembly is so obvious that you may wonder why somebody did not have this idea long ago.

I did find this on the Femto GitHub and got a chuckle. Yes, dear author, that is exactly the piece of information I want explained. Too funny.

replies(1): >>44383234 #
4. dmkolobov ◴[25 Jun 25 21:09 UTC] No.44381862[source]▶
If I had to guess:

1. lack of native GC, you had to roll your own by providing a runtime.

2. lack of tail-call elimination in V1 of the spec. This essentially forces you to trampoline everything.

replies(1): >>44382157 #
5. stassats ◴[25 Jun 25 21:55 UTC] No.44382157[source]▶
> lack of tail-call elimination

But lisp doesn't need TCO.

replies(2): >>44383319 #>>44384522 #
6. no_wizard ◴[26 Jun 25 00:55 UTC] No.44383234{3}[source]▶
It doesn’t have to very last trapping of a proper LISP if I recall correctly is why I say kinda.

It’s an S expression language though with some similarities to LISP

replies(1): >>44383311 #
7. bevr1337 ◴[26 Jun 25 01:06 UTC] No.44383311{4}[source]▶
Thanks for clarifying. I've been speculating on this for a while but haven't had the chops to know what to research. The details folks are providing are yielding much better search results.
8. bevr1337 ◴[26 Jun 25 01:08 UTC] No.44383319{3}[source]▶
Whoa, what a surprising fact! I had not considered TCO in LISP was "nice to have." That's a good example where it's easy to hack but hard to make production ready.
9. dreamcompiler ◴[26 Jun 25 05:47 UTC] No.44384522{3}[source]▶
Common Lisp doesn't technically require TCO but Scheme does.

IMHO a Common Lisp without TCO is not a serious tool because I cannot write tail-recursive code with it.

replies(1): >>44387509 #
10. wavemode ◴[26 Jun 25 13:59 UTC] No.44387509{4}[source]▶
Typical programming style in Common Lisp is procedural, not functional.

The CL spec doesn't guarantee tail-call elimination, and most CL implementations don't guarantee it either (it could be on or off depending on the current optimization level). So most people don't bother relying on it.