Do you want to use Functional Programming on 96KB RAMs device? | Hacker News Reader

Do you want to use Functional Programming on 96KB RAMs device?

lambdachip.com

39 pointsnalaginrut10 comments

I like the idea, however when discussing about memory sizes for functional languages, we should recall that Lisp was developed in the 50's and was running on hardware like the IBM 401.

So 96KB for 50's hardware would already be quite good.

Functional programming with tail calls and potentially even tail call allocation (like in the ATS language) can have C (or greater in the case of potential compiler hints) re performance/memory usage. So it's not exactly surprising.

That being said, this is pretty cool. As far as the current landscape of embedded languages go, I'd take a functional lisp over C, probably, but I'd have to dive deeper on more extensive examples and look into how easy it is to map other boards/devices.

what are the chances of seeing more language-enabled boards, javascript, python boards? if they were designed and built around a particular language it could be efficient

Can the chips be self-hosting? By which I mean can a lambda chip compile code to run on another lambda chip?

It's a really interesting and challenging question. I've written a boostraping compiler, and I enjoyed the developement. It's a strange and sleepless process.

Well, back to the topic, maybe we can challenge it for LambdaChip someday in the future. ;-)

A Cortex board that runs Lisp, not the reincarnation of the Lisp machine. May be useful for some cases, but embedded and FP are not usually comfortable partners.

Yes, it's a neglected topic, but it's the topic that LambdaChip wants to try.

I do functionalish programming on a 400KB RAM device using eLua and Wasm.

Lua is basically a Scheme with a different syntax.

Agreed, we actually have a Lua-5.2 frontend underway. We may officially support Lua in the future, but for now, we need to support full r7rs in Scheme.

I am glad you have picked up the embedded functional flame. I assume when you say R7RS, you mean R7RS Small [1], Shen [2] might be a better fit for embedded dev as it is typed and has multiple backends including Scheme [3]

I think things get interesting when we have a core or more per pin and hardware GC.

[1] https://small.r7rs.org/

[2] https://shen-language.github.io/

[3] https://github.com/tizoc/shen-scheme

Reply on news.ycombinator.com