Hacker News Reader

Comment by rrmm | Hacker News Reader

I'm interested in this too. What I've been wanting is a single chip solution.

I've designed various systems based on arm7 chips, but my naive lisp implementations barely fit any usable system into the onboard RAM (~90k).

I've been waiting for atmel or someone to put something like 1MB or so on chip, but the market for such devices isn't there.

FORTH seems better able to fit and has a little better fundamental underpinnings than BASIC.

View on HN

The STM32F429 Discovery board [1] has 2MB of Flash and 256KB of internal RAM. It also looks like it has 64 Mb of SDRAM which might be what your looking for. We have an engineer where I work who is trying to get mRuby up and running on one. It has tons of memory for a Cortex-M part/board

[1] - http://www.st.com/web/catalog/tools/FM116/SC959/SS1532/PF259...

That's definitely more like it. 256k is a reasonable chunk. Although at $20 and 144LQFP, it's a pricey and kind of chunky part. It exists though which is nice.

How does mRuby do GC; does it ref count? I don't think lisps could really afford to be ref counted, unless you pool the cons cells into groups that are ref counted. My implementation was a dual semispace collector which didn't help matters wrt space.

Amusingly, I've built industrial controllers using the 8052AH-BASIC chip, but it was pretty bare bones.

I've used those too. Having to wire up the 'LS373 latch was always annoying though.

Reply on news.ycombinator.com