PunyForth: Forth inspired programming language for the ESP8266 (opens in new tab)

(github.com)

88 pointsdentrado9y ago45 comments

45 comments

I'm reminded of my good old Forth times (6502 on Commodore PET), and I'm glad that Forth gets the increasing attention which it actually deserves.

I don't know if there is any other programming environment at all where you can have a extensible programming language, editor, compiler, interpreter, debugger, libs, and interactive shell in a few KB.

shakna9y ago

As much as I like Forth, it isn't the only player in the restricted memory arena. Several models of the Commodore PET came with 32kb of RAM, and 18kb of ROM. Here are some that reflect similar requirements.

uLisp [0], 2kb of RAM, and 32kb to store the interpreter in. You can write programs at the REPL, and burn them into the EEPROM.

TinyBASIC [1], since dead, is a port of the version of BASIC that used to run on the Altair 8800, which was an even more constrained system than most PETs! Altair BASIC could run on 4kb of RAM.

Python-On-A-Chip [2], is a subset of Python. At a guess, from the supported platforms, it can run on around 2kb of RAM.

iArduino [3], is a C interpreter that can run in 2kb of RAM.

MicroPython is looking to come to the ESP8266 [4], which is much, much bigger at 512kb. Which basically makes it the fattest of these, and impossible to run on something like the PET.

Bitlash is a completely new language [5], and can run on 2kb of RAM, and takes up surprisingly little storage space.

All of these languages offer at least a REPL, as well as compatibility with a wide range of libraries, and most have at least traceback and exception support.

This is not to say Forth is a bad choice! It works great in tiny memory, with better support for a full language specification than many other languages.

But not everyone wants a Forth... And there are plenty of choices out there.

[0] http://www.ulisp.com/

[1] http://hackaday.com/2011/08/28/basic-programming-on-an-ardui...

[2] https://github.com/jsnyder/python-on-a-chip

[3] http://n.mtng.org/ele/arduino/iarduino.html

[4] https://github.com/micropython/micropython/tree/master/esp82...

[5] https://github.com/billroy/bitlash/wiki

progman9y ago

Yes, there are plenty of options. However, most of those options lack something while Forth is all-inclusive. All your options have no compiler while Forth compiles to threaded code which is pretty fast, about 1/10 the speed of assembler according to my experience. Your options have a linear space consumption while Forth code compiles "logarithmic" due to its extensible vocabulary. That means, if you have an 8K system (like on my former PET-2001) you need 4K for a convenient basic Forth for instance, while you can put a lot more stuff in the other 4K, due to the threaded code.

2 more replies

pjmlp9y ago

Additionally one should add that the mainframes where Lisp was initially developed, were quite constrained, even when compared with 8 bit micros.

tines9y ago

This is fantastic! Forth's simplicity is a great fit for exploratory programming on resource-constrained systems, and I'm a fan of anything homoiconic.

Maybe one day I'll get my dream of having a forth or lisp with the type system of haskell.

vanderZwan9y ago

Have you looked at Kitten? It's statically typed[0], and a descendant of Cat[1], which is a descendant of Joy[2]. So those might also be interesting languages to investigate

> The Kitten Programming Language

> Kitten is a statically typed, stack-based functional programming language designed to be simple and fast. It is a concatenative language, combining aspects of imperative and pure functional programming. There is an introduction available and a tutorial in progress.

http://kittenlang.org/

[0] http://evincarofautumn.blogspot.se/search/label/Concatenativ...

[1] http://www.codecommit.com/blog/cat/the-joy-of-concatenative-...

http://www.codecommit.com/blog/cat/the-joy-of-concatenative-...

[2] https://en.wikipedia.org/wiki/Joy_(programming_language)

eggy9y ago

Maybe Shen would work for you? [1] A Lisp with optional static type checking, optional lazy evaluation, pattern matching, a built-in Prolog, and more.

[1] http://www.shenlanguage.org/

jstimpfle9y ago

Aren't these conflicting ideas? To me a big advantage of dynamic type systems is that specifying behaviour and specifying what parts of the behaviour are statically computed are separate tasks.

david-given9y ago

There are some strongly typed Forths --- although I've never seen one that was out of prototype. Here's one I had a brief look at:

http://www.arestlessmind.org/2009/02/03/intro.html

Of course, since Forths are built around a REPL, to do type checking you need to keep the type objects around at run time, so it may not be suitable for very small systems.

tines9y ago

Yes, but they are not conflicting in the sense that they are competing to serve the same purpose.

For exploratory programming, dynamic type systems are nice. For building reliable long-running systems or ones where failure is extremely costly, I like static.

thurn9y ago

Yes. For example, the powerful "transducers" concept from Clojure essentially cannot be represented in the Haskell type system.

1 more reply

nine_k9y ago

Did you see the Factor language?

throwaway76459y ago

Factor is cool, but never really took off or got much attention. The creator, Slava has moved on to other things (Swift compiler I think). I posted something on another concatenative language (kitten) on here recently and the author jumped on and the conversation was good (Factor was brought up as well).

reikonomusha9y ago

This is great work. I was expecting a toy Forth, but this one is pretty serious. I was happy to see quotations, immediate words, and Factor-style combinators.

afeezaziz9y ago

I am wondering why the team from mbed has not made an attempt to make ESP8266 compliant with mbed. Yes, I understand that the architecture is totally different but just that I like mbed so much so that I want to see it used in more hardware.

jakobdabo9y ago

The "team from mbed" is employed by ARM.

kbumsik9y ago

mbed team would not do that. The compiler used in mbed platform is NOT gcc but ARM Compiler.

jakobdabo9y ago

Great work indeed. Does it support the new ESP32 processors?

aseipp9y ago

The ESP32 also uses an Xtensa, the LX6, which should have the same basic instruction encoding, etc (and I think they both use FreeRTOS, as well, but I only have an ESP32). Probably shouldn't be too hard. You can also stuff more functionality into it, too.[1]

[1] Related: I can't actually even find the documentation on what's precisely changed between the LX6 (ESP32) and the LX106 (ESP8266), other than the basics on the datasheets - I don't know of any documentation on e.g. any different architectural details. Maybe you have to ask Espressif. Or Cadence. Anyone have any idea?

jakobdabo9y ago

Their datasheets only mention the PIN definitions and electrical characteristics of the microcontrollers. Try to get the information from the reference manual[0]. Also there are these books[1][2] by Neil Kolban.

[0] https://www.espressif.com/sites/default/files/documentation/...

[1] https://leanpub.com/ESP8266_ESP32

[2] https://leanpub.com/kolban-ESP32

edit:

More information about the differences: http://www.cnx-software.com/2016/03/25/esp8266-and-esp32-dif...

1 more reply

MrBuddyCasino9y ago

Most importantly, the ESP32 has register windows.

1 more reply

stavros9y ago

Why can't we get Rust for the ESP8266? :( I really dislike C, I'd love a higher-level language. As great as MicroPython is, the interpreter needs too much RAM, and I think Rust would be a great alternative.

semi-extrinsic9y ago

What's wrong with Lua? It has first class support thanks to NodeMCU, you can customise your firmware builds to be lean on resources using a dead simple online tool, and it even lets you run code on the ESP8266 in an interactive REPL environment.

StavrosK9y ago

Lua is strictly worse than MicroPython, I find. Memory usage is worse, and managing programs on it is bad too, although I haven't used it as much as MicroPython.

1 more reply

pjmlp9y ago

Because no one has yet bothered to port it?

As a side note, there are quite a few options to do safe micro controller system programming for those that prefer AOT compiled languages without C like unsafety.

http://www.mikroe.com/mikropascal/

http://www.mikroe.com/mikrobasic

http://www.astrobe.com

http://playground.arduino.cc/Code/AVR-Ada

progman9y ago

Wouldn't Rust be overkill?

I think Forth is much more suitable here. You can implement a lot of features in 512K. You can extend the code (and language!) at runtime, you can debug remotely, and do other nice things.

bitwize9y ago

Yes, but Forth doesn't have a borrow checker, so how will you know you're using memory safely?

1 more reply

dbcurtis9y ago

Isnt there an axpetimental Rust backend for ARM Cortex? ISTR seeing it discussed here.

masklinn9y ago

Rust has Tier 2 support[0] for ARMv6 to ARMv8 and Tier 3 support[1], so there are experimental limited[2] backends for the various Cortex in Tier 3.

However the ESP8266 is built around Tensilica's Xtensa, not ARM's Cortex. There currently is no support whatsoever for Xtensa in LLVM (let alone in Rust).

Current microcontroller work is on MSP430[3] with work being planned for Amtel AVR[4] since the AVR branch was merged into LLVM in November 2016. Xtensa is quite literally nowhere[5].

[0] guaranteed to build & binaries provided but tests are not automatically run

[1] supported in the codebase but no guarantees that it even builds

[2] only Core, no standard library

[3] https://github.com/rust-embedded/rfcs/issues/20

[4] https://github.com/rust-embedded/rfcs/issues/3

[5] https://github.com/rust-embedded/rfcs/issues/2

1 more reply

NateyJay9y ago

The ESP8266 uses an Xtensa core, not ARM.

j / k navigate · click thread line to collapse