Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

uLisp – Lisp for the Arduino

ulisp.com

Post reply on HN
Print view View original

21–30 of 34 posts

Re: uLisp – Lisp for the Arduino

#21

Post by chipsy »

pcl wrote: post #7

uLisp includes a mark and sweep garbage collector. Garbage collection takes under 1 msec on an Arduino Uno or under 3 msec on an Arduino Mega 2560. I wonder how predictable and controllable that is. Adding GC to what is likely to be a real-time device seems like a potential showstopper. But if the numbers can be a bit more firm, or if the application can receive alerts on pause / resume, that might not be such a big…

In a real-time application that has to run on constrained hardware, the most interesting use of a GC language is as part of a compilation/configuration process for the actual runtime code or assets.

This is the approach of e.g. Naughty Dog's old GOAL language: it isn't doing Lispy things in the engine code, so much as using Lisp as a very powerful macroassembler. This approach is potentially advantageous over having a batch process to compile and upload to the device since it presents some important groundwork for doing "live editing". The performance of the shipped application is effectively unlimited, since the user code has so much control over the resulting output.

The main downside is that the resulting app isn't designed for portability, being so oriented towards the machine code rather than an abstract layer: Getting the portability and the live-editing functionality and the performance is a considerably harder problem to visualize, but production C++ game engines are doing it now.

Re: uLisp – Lisp for the Arduino

#22

Post by lisper »

sigil wrote: post #20

This looks really neat! There was a story about Lisp at JPL that got me dreaming about embedded REPLs: > The Remote Agent software, running on a custom port of Harlequin Common Lisp, flew aboard Deep Space 1 (DS1), the first mission of NASA's New Millennium program. Remote Agent controlled DS1 for two days in May of 1999. During that time we were able to debug and fix a race condition that had not shown up during gro…

I have TinyScheme running on a STM32F415.

https://sc4.us/hsm/

Re: uLisp – Lisp for the Arduino

#23

Post by sigil »

lisper wrote: post #22
sigil wrote: post #20

This looks really neat! There was a story about Lisp at JPL that got me dreaming about embedded REPLs: > The Remote Agent software, running on a custom port of Harlequin Common Lisp, flew aboard Deep Space 1 (DS1), the first mission of NASA's New Millennium program. Remote Agent controlled DS1 for two days in May of 1999. During that time we were able to debug and fix a race condition that had not shown up during gro…

I have TinyScheme running on a STM32F415. https://sc4.us/hsm/

I love TinyScheme. I love NaCl. This is wonderful!

Re: uLisp – Lisp for the Arduino

#24

Post by TeMPOraL »

AceJohnny2 wrote: post #3

The AVR cpu at the core of the Arduino uses a (modified) Harvard architecture., which separates code from data memory. A key feature of lisp of that code IS data. How does uLisp bridge the contradiction?

Code IS data, that's a fundamental fact of reality, and not something mere chip architecture can invalidate. Separating "code memory from "data memory" only means you can't have bits from "data memory" flowing directly into the chip to be run. It doesn't prevent you from having some code in "code memory" that would translate the data into code, by e.g. means of flow control.

Re: uLisp – Lisp for the Arduino

#25

Post by TeMPOraL »

Steltek wrote: post #11
AceJohnny2 wrote: post #3

The AVR cpu at the core of the Arduino uses a (modified) Harvard architecture., which separates code from data memory. A key feature of lisp of that code IS data. How does uLisp bridge the contradiction?

> A key feature of lisp of that code IS data. I think this is an oversimplification. "Code is data" means several things: 1. Both use the same ASCII representation to humans (homoiconity) 2. Lisp Macros operate using data traversal functions 3. You can load and eval code on the fly Really only #3 is invalidated under Arduino constraints. #1 is the source at rest and #2 is compile-time.

3. is not invalidated under Arduino in a general sense. The split into "code memory" and "data memory" is only relevant to the meaning of "code" and "data" assigned by uC-level assembly. There's nothing stopping your code to reinterpret data as code. Consider: that's how interpreters and VMs are made :).

Moreover, 1-3 together are more than sum of its parts. The idea that "in Lisp code = data" doesn't mean that elsewhere it doesn't; code = data is a fundamental truth about the nature of information (that many other languages and their ecosystems did quite a lot of work to obscure). Lisp only makes exploiting that truth much easier than other languages.

Also, a nitpick. #2 is not "compile-time". Macros can and are expanded at what you'd call "run-time" as well. Generally, Lisp systems don't have "compile-time" and "run-time" you know from other languages; the split is usually read- / load- / compile- / run-time, with some other "times" sprinkled in (e.g. "macroexpansion-time"), and this split is not necessarily sequential - you can read/load/compile stuff in what would usually be called "runtime", and read/load/compile phases have full access to features of the running Lisp image, including changes you made previously - so, for example, code you're just compiling can, during its read-time, refer to variables you previously created at run-time and run computations on them.

Re: uLisp – Lisp for the Arduino

#26

Post by TeMPOraL »

askvictor wrote: post #13
WillPostForFood wrote: post #6

Lisp for Arduino sounds great. The C programming barrier has kept me from tinkering with Arduino. Does this give you access to arduino shields/expansion boards (wifi/gps/screens), or will that be dependent on device drivers?

Check out esp8266/nodemcu. They can be programmed in lua or micropython (or arduino ide). Not as much of an ecosystem/community, but growing. Includes wifi on-chip, and dirt cheap.

There's also a Lisp for ESP8266 under development :). See:

https://github.com/yesco/esp-lisp

Re: uLisp – Lisp for the Arduino

#28

Post by genop »

lisper wrote: post #22
sigil wrote: post #20

This looks really neat! There was a story about Lisp at JPL that got me dreaming about embedded REPLs: > The Remote Agent software, running on a custom port of Harlequin Common Lisp, flew aboard Deep Space 1 (DS1), the first mission of NASA's New Millennium program. Remote Agent controlled DS1 for two days in May of 1999. During that time we were able to debug and fix a race condition that had not shown up during gro…

I have TinyScheme running on a STM32F415. https://sc4.us/hsm/

Any BSD kernel drivers for this?

Re: uLisp – Lisp for the Arduino

#29

Post by pjmlp »

Volt wrote: post #16
nice_byte wrote: post #2

> It's also an ideal language for expressing complex ideas, such as [...] finding the shortest route on a map So, I googled "Dijkstra's Algorithm in Lisp" and got this: http://richardsherriff.com/?p=233 Now, I'm no lisp expert, so I can't judge whether the author of this code actually knows what they're doing, but I know for sure that in an imperative language the implementation of the algorithm is much more concise…

>in an imperative language Lisp is capable of imperative programming.

I would say multi-paradigm instead.

Re: uLisp – Lisp for the Arduino

#30

Post by kazinator »

Though uLisp has tail-call optimization, if the C compiler doesn't, then the garbage collector needs stack in proportion to list length:

  void markobject (object *obj) {
    if (obj == NULL) return;
  
    object* arg = car(obj);
    if (marked(obj)) return;

    int type = obj->type;
    mark(obj);
  
    if (type != SYMBOL && type != NUMBER) { // cons
      markobject(arg);
      markobject(cdr(obj));   //
It's simple enough to obj = cdr(obj) and wrap a loop around this.
Post reply on HN