The Byte magazine referenced in the article:
Byte Magazine Volume 04 Number 08 - LISP
I certainly appreciate the theoretical beauty of lambda calculus, but I do wonder what the point is from a modern implementation perspective.
Most hobby Lisps I run into seem to be more of a lambda calculus rite of passage than aiming for practical use.
I've implemented several Lisp interpreters myself in different languages, very much focused on practical use, and they don't look anything like this.
An argument from Justine Tunney's LC implementation [1]:
> Programs written in the binary lambda calculus are outrageously small. For example, its metacircular evaluator is 232 bits. [...] Something like this could have practical applications in compression formats, which need a small busy beaver that can produce large amounts of data. It's also just downright cool.
> For example, if you built a compression tool you could have it encode a file as a lambda expression that generates it. Since it's difficult to introduce a new compression format that most people haven't installed, you could prefix the compressed file with this 383 byte interpreter to get autonomous self-extracting archives that anyone can use.
I partway expect @tromp to show up in this thread, but see his excellent site[2] for discussion of the algorithmic information theoretical applications of LC, exploration of Kolmogorov complexity, etc.
1. https://justine.lol/lambda/
2. https://tromp.github.io/cl/cl.html
Every time I read anything by jart, I am reminded that there is a level of intellect so far above most that it doesn’t even make sense.
“I shall write a brainfuck interpreter in Lambda Calculus for fun.” Sure, sure, as one does.
jart is on a whole other level. I love the seeing the stuff she puts out and the writings she does. So, so, fun to be a jaw-dropped onlooker. Not to mention the tools are genuinely useful and incredibly cool. Llamafiles, alone, should win some kind of AI accessibility award.
Very interesting, thank you!
Downvotes?
HN never ceases to surprise me!
I remember having memorised lambda calculus solving techniques to pass my CS exams on paper where the 2-forearm wide lines had to be wrapped multiple times.
And now, just like a decade ago, my eyes glaze over when I read about this.
I can easily reason about functions that return functions that return functions but this just completely goes over my head.