Hacker News new | past | comments | ask | show | jobs | submit login
Pseudo Scheme: Scheme Implemented on Top of Common Lisp (cmu.edu)
52 points by interpunct 3 months ago | hide | past | favorite | 17 comments



Like almost always, there are chapters about that in PAIP:

22 Scheme: An Uncommon Lisp[what a pun!] https://github.com/norvig/paip-lisp/blob/main/docs/chapter22...

23 Compiling Lisp [Scheme] https://github.com/norvig/paip-lisp/blob/main/docs/chapter23...


I feel like one of the major benefits of using Scheme is how easy it is to interoperate with just about any environment. Want to embed Scheme in your C application? Take your pick from GNU Guile, Chibi, Gambit, or Chicken. How about the JVM? Kawa is solid, option. .NET? IronScheme has got you covered. Javascript? There is LIPS or BiwaScheme.


I agree. I recently started using an R4RS Scheme called "STk", the precursor to STklos, which I modded into a build system (replacing Make and Sh, for the most part) and a graphics DSL front-end to C. The author wrote a demo app web browser with Scheme as a scripting language, as was the original plan with Mozilla, and Tk as the GUI. Writing extensions in C is a breeze too, as it supports dynamic loading:

https://github.com/egallesio/STk

I like Common Lisp too, although not as much as Scheme day-to-day--it is nice now to have both. I was thinking about adding a TinyCLOS using Common Lisp. Pseudo Scheme transpiles source to CL, so I might be able to do it with CLOS underlying it instead of how TinyCLOS does it.

Jonathan Rees, the original author, has a Pseudo Scheme repo up on Github, with a branch that uses ASDF (which I am testing with SBCL).

https://github.com/jar398/pseudoscheme

Thanks for your comment, and thanks to the maintainers of both of the linked Scheme implementations.


> Pseudoscheme consists primarily of a Scheme to Common Lisp translator that is written in Scheme. To obtain a version of the translator that runs in Common Lisp, it is applied to itself.

Yes!


In the CL world: to embed CL in your C app: ECL; interface with the JVM: ABCL and LispWorks; for .NET: Bike (https://github.com/Lovesan/bike/). JS: JSCL and Valtan (might need more love).


CLASP in C and C++ via LLVM.

LispWorks and Allegro CL as shared libraries in various environments. LispWorks in iOS and Android Apps.

SBCL, I think, now possibly works also as a shared library.


I've sometimes done the opposite - when working in scheme/racket, searched for implementations of things like loop, defmacro or tagbody. (They have their quirks, but when they're a good fit for a problem they can save a lot of typing.)


Chicken has a Common Lisp style `loop` egg: https://wiki.call-cc.org/eggref/5/loop I think it's based on a more generic Scheme version but I don't have a link to the original.

SLIB has an implementation of pretty much the entirety of CL's `format` that I ported to Racket: https://docs.racket-lang.org/slib-format/index.html

I also implemented a bunch of other CL stuff for Racket in https://docs.racket-lang.org/soup-lib/index.html (but not `tagbody`. Yet.)


Oops! Chicken is BSD licensed, but this is GPL.

If you ship a program that includes this, it is tainted.

Possibly, the code generated by the loop macro doesn't fall under the license, but you still have to ensure that the macro itself is scrubbed from your application.


Or... Just release your program under the GPL.


Defmacro is trivial to implement in syntax-case, but you are in for a world of pain if you want to use bindings across modules.

In have an implementation of tagbody for guile scheme somewhere, using delimited continuations. I have seen it for racket as well. There is an implementation of loop for racket, but it relies heavily on set! which will tank performance, since chez and racket (and many schemes) is pretty bad at not boxing mutable values.

I wrote this which is definitely not as powerful as loop, but definitely more powerful than rackets for loops: https://git.sr.ht/~bjoli/goof-loop


Defmacro isn't hygienic, while syntax- is. Is there way to use arbitrary names in hygienic syntax- macros?


If you know the name ahead of time, syntax parameters[1] (if your scheme supports them) work well. syntax-case lets you break hygiene to insert new names in a controlled fashion (but has issues with names known ahead of time; see linked paper). Renaming macros do too, I believe.

1: http://scheme2011.ucombinator.org/papers/Barzilay2011.pdf


syntax-case can introduce arbitrary bindings using datum->syntax.


I'm curious how it handles call/cc. I at least think you would have to make your own continuation stack, unless there's some way to get at the internals of unwind-protect that would make doing it natively possible.


That's how PAIP does it, their interpreter uses continuation passing style:

22.5 An Interpreter Supporting Call/cc

https://github.com/norvig/paip-lisp/blob/main/docs/chapter22...

23.2 Introducing Call/cc [see the next chapter, 23.3 The Abstract Machine, for the implementation in bytecode]

https://github.com/norvig/paip-lisp/blob/main/docs/chapter23...


Has someone ported this to Emacs?




Consider applying for YC's W25 batch! Applications are open till Nov 12.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: