Hacker News Reader

Comment by hoprocker | Hacker News Reader

Thanks so much! I was really stoked by yesterday's related post (and your discussion in the comments). I also find it really interesting to see Scheme used in production; I sort of marginalized it (unfairly to myself, I'm sure) after college as being a good toy Lisp for things like demonstrations in SICP.

View on HN

I sometimes wish there was a language inbetween scheme and common lisp that actually caught on. They're both powerful languages, but to me the little scheme I've done makes me feel like I'm just reinventing common lisp as I go along, and sometimes common lisp leaves me wanting for something a bit "smaller" that doesn't take up 30mb of memory just to open the REPL. This especially holds true in the mobile world.

Apart from the syntax and our current lack of a REPL outside of Windows, Dylan (http://opendylan.org/) sort of fits in that space. We're a lot like a slimmed down Common Lisp (designed by some of the same people that standardized CL), and bigger than a Scheme.

We're currently working on a lot of stuff, including an Objective C / Dylan bridge, a new build toolchain and an experimental REPL that may work outside of Windows until we can bring over the real one (once our LLVM backend is in place).

I keep hearing about Dylan (good things) and have checked out the site a few times, but never sat down and actually learned the syntax and built something with it. I'll try and give this a go in the next few days. Is it like CL where there's a language spec and independent implementations?

Thanks for the tip!

We have a specification: http://opendylan.org/books/drm/

We have, in the past had multiple implementations, but due to community size, desire to evolve things, etc, we narrowed it down to Open Dylan being supported and maintained.

It's true that CL's huge feature set promotes implementation bloat, but that doesn't completely rule out having a compact implementation.

In fact, mocl can run with only a few megabytes of memory, and has options to control the heap size.

If this seems unbelievable, feel free to stop by the mocl office in Redwood City, and I will explain it :)

Funny, I'll be in Redwood City tomorrow. Didn't know you guys were based there. Question for you: did you write your own implementation, or are you adapting something like ECL? Do you have any articles posted on how it works?

mocl derives from some past academic work, although not ECL. In spite of this, it is very much a new and unique implementation, and I have sunk years of dev time into making it a reality.

I may or may not be in the office Sunday... feel free to reach out (call/text the number on wukix.com).

The talk I gave at ECLM 2013, in Madrid, may be informative. http://www.youtube.com/watch?v=l44kiFp1EJ8

Clozure Common Lisp started 1987 as Coral Common Lisp on a MAc with 5 MB memory on a 68000 CPU with 8 Mhz. I use it on ARM boards. For example I have a quad core ARM board which is computationally equivalent of some of the newer Samsung phones - Clozure CL runs perfectly fine on those. 100+ times faster than Coral Common Lisp on the early Macs.

That's awesome, didn't know it ran so well on smaller systems. Are there certain settings you start it with? I usually just start with the default parameters and haven't explored heap sizes and all that.

Not sure what you mean by between Scheme and CL, but if there's one Lisp that'll catch on to today's practical needs, it's probably Clojure (and ClojureScript).

What I meant by inbetween is something exactly like CL with a smaller footprint and possibly a smaller library. I believe CL more than covers today's practical needs (real threading, compiling to machine code, calling out to C, etc etc). I might choose Clojure if I was writing something highly concurrent that required a more functional approach, but for any other general purpose app I'll take CL any day. (Disclosure: I've never actually built anything in Clojure).

No its not. I wonder where people get these ideas...

Reply on news.ycombinator.com