Then I let my coworkers graffiti the slides with the same lisp as I gave my talk, as it was an open web server where you submitted them. A tiny demo is left running at https://www.adhocteam.club/ , but I should write it all up.
I don't want to take away from this, but any time a home built lisp interpreter comes up with a "let me show you how it works", I always have to reference back to the MAL (make a lisp) project:
I find it's better to write it yourself than just have it explained to you. It really doesn't take that long to do, and it's broken into nice chunks of discrete work.
I love these articles (plus the "make a lisp" project as well). They give me a lot of hope in the future: Next time a customer insists that its software should be delivered in Java, i will first write my own informal, bug-ridden implementation of Lisp on top of it, and write the rest of the software in such Lisp.
I'm unclear about function invocation here... FTA:
A function invocation. This comprises a list where the first element
is the function and the rest of the elements are the arguments. first
takes one argument, (1 2), and returns 1.
My problem is, the same way in every lisp I've seen, (1 2) qualifies as a function invocation. I'm not sure how the interpreter would know not to invoke 1 on 2, in these examples:
In most (all?) lisps that's actually exactly what would happen. It would try to call the function '1' (and throw an error since 1 is not a valid function). If you want to have a list you'd typically have something like (list 1 2) or you could quote it '(1 2).
In this case it seems to actually check if the first element of the list is a function and only treat it as a function call in that case.
but you're not passing it the argument '(1 2). Lisps are usually (always?) strict evaluation, so the form `(list 1 2)` or `(quote (1 2))` would return the form (1 2). First applied to this form would return 1. The specification is correct that first applied to (1 2) would return 1. The fact that this means most invocations would look like `(first '(1 2))` is irrelevant since the argument to first here is not '(1 2) but is actually (1 2).
You are actually in violent agreement with the parent.
The grandparent poster was confused why the syntax (foo (1 2)) can be used to apply FOO to (1 2). As the parent points out, for typical Lisps this would actually give an error; instead, (foo '(1 2)) would be the appropriate syntax to apply FOO to the form (1 2). Indeed, when strictly evaluating (foo '(1 2)), first the arguments are evaluated. Since functions self-evaluate, FOO evaluates to itself, while '(1 2) evaluates to (1 2). Then, FOO is applied to (1 2). This is in complete agreement with what you said and what the specification says.
However, the Lisp interpreter at hand actually self-evaluates lists whose head is not a function. Thus (1 2) self-evaluates and (foo (1 2)) has the same effect as (foo '(1 2)).
> Since functions self-evaluate, FOO evaluates to itself
No, in Lisp FOO is a name of a function, not a function itself.
Thus FOO evaluates to a function (otherwise it would be an error) in a Lisp-1 like Scheme. In a Lisp-2 like Common Lisp, one would say that the function value of FOO is retrieved.
> Indeed, when strictly evaluating (foo '(1 2)), first the arguments are evaluated
Actually not. In Lisp the first item needs to be looked at first. If it is determined to name a function, then we can evaluate the arguments, of which there is only one in this case.
Fantastic comment, this cleared up one or two headscratchers I encountered while patterning a Lisp I was writing in Lua with LPeg (I had just encountered LPeg and wanted to do something fun with it) on Peter Norvig's Lis.py. It didn't 100% square with what I recalled from other Lisps.
Or in other words, a list is interpreted as a call if it starts with a function, or else it's interpreted as just the list as data. (1 2) is the latter case, since 1 is not a function.
It seems almost a right of passage for a programmer to build/attempt a lisp/pseudo-lisp implementation. Quite honestly, when I made my (very poor) attempt, it was one of the most fun, and most frustrating projects I had done at the time, and I would highly recommend every developer have shot at building some kind of lisp.
One thing I did notice in this implementation, is the lack of quoting and cons notation. Although, cons is a weird one in the lisp world given it is the only infix operator in the language.
Then, instead of adding variables, I wrapped most of the functions on a canvas in the env and used it to generate my talk's slides: https://github.com/llimllib/adhocteam.club/blob/master/slide...
Then I let my coworkers graffiti the slides with the same lisp as I gave my talk, as it was an open web server where you submitted them. A tiny demo is left running at https://www.adhocteam.club/ , but I should write it all up.