A great example of the rule is that C++ rediscovers car/cdr at a glacial pace in the template language. In C++26 one can finally get the car of a typename parameter pack with Args...[0].

I've no idea why they don't introduce car/cdr functions and nil for empty parameter packs and allow to store parameter packs instead of the current syntax insanity.

In a struct! Pseudo code for extracting the types from subclasses and calling a multimethod on them:

  template 
  struct Foo : Visitor {
    Result res;
    UntypedList lst; // This does not exist!
    Tail... tail;    // This is not possible!
    Foo(UntypedList l; Head hd, Tail... tl) : lst(l), tail(tl) {
      hd->accept(*this);
    }
    void visit(float *f) {
      res = Foo(append(lst, f), tail)::res; }
    }
  };
 
  // Base case that calls the multimethod omitted.

There are two issues here: You can't store the parameter pack in Foo() which is later required by the continuation in visit(). And you would have to use tuples and tuple_cat() instead of UntypedList.

Why can the compiler not infer the types in such an UntypedList? Why is there no car/cdr for tuples?

I conjecture the line is not so easy to draw.

If you are creating Lisp because you want to create Lisp, like creating Lisp, want to show off creating Lisp, that obviously is not what the Law is about.

Furthermore, if you create Lisp because you know the Law, know it is inevitable, and want to avoid the caveats and missed bars by doing so explicitly, well then that also is not what the Law is about.

But if you are going about your business, focused on your independent goal, realize you need Lisp like lists, and then 250 lines of code later realize you have created a solid Lisp unintentionally? Well, congrats on falling into the trap but not getting hurt!

—

Personally, I have created both Lisp and Forth multiple times for suitable projects where I wanted some flexible runtime scripting. I am not familiar with the standard version libraries of either and don’t need them.

Minimal foundation implementations are extremely easy to create, and eliminate any dependencies or sources of mystery.

Anyone know of any other well designed mininal languages?

"for any sufficiently X the Y" just means that if you don't observe Y right now, just increase the magnitude of X and you'll inevitably reach the conditions for Y to happen.

Famous use of that pattern is Arthur's Clarke "Any sufficiently advanced technology is indistinguishable from magic.". As with OP's quote, this is also vague and imprecise but the point of the idea is that there is a gradient towards which advancements in technology bring us towards a situation where the average person no longer understands how it works and it may as well be magic (while not necessarily defining exactly when does that transition happen)

No, I think it's pretty fair. One could argue about these, but except for "slow" these are more quality qualifiers, rather than quantity. So you either agree it's "bug-ridden" or not (i.e. the number and seriousness of bugs in it is negligible by whatever standards). And I think even "slow" can be discussed in the same manner, the actual speed is quantitative, of course, but in the end one either argues that this speed is "slow" or isn't. So, given some rhetoric skills of your opponent it's at least possible for that statement to be proven false, if he convinces you the implementation actually isn't slow nor bug-ridden, or at least if there's no Lisp-implementation indeed.

But what is "sufficiently" complicated? Now it's a silly statement that just doesn't pass Popper criterion: even if nobody dares to deny your program is complicated, one can always say it isn't sufficiently complicated, hence the fact it doesn't contain Lisp-interpreter disproves nothing. A man of culture wouldn't use such definitions.

Isn’t that sort of the point? It makes the argument impossible to dispute. If it’s not slow or big-ridden, then it’s just not sufficiently complicated. Which you’re correct to call out.

I think what makes it work is that Common Lisp very quickly becomes a faster way to do things because of how insanely efficient it is.

The problem with that is that there are spaces between map, -, and '(1 2 3). The only way to get spaces into a name is by using vertical bars:

  (defvar |do i 10| 1.100)

It's almost but not quite arbitrary. It still has to be a valid Rust token stream with matching braces/parens. Since it's whitespace insensitive with respect to tokens and "-" is its own token, "(foo-bar)" and "(foo - bar)" result in the same token stream minus the span information.

You can use proc_macro::Span.line()/column() [1] to track how much whitespace there is between tokens and merge them, but the macro will have to rename them to valid Rust identities ("foo-bar" to "foo_bar") and make sure there's no collisions.

[1] https://doc.rust-lang.org/proc_macro/struct.Span.html#method...

Some Lisp compilers like SBCL are already capable of more extensive compile-time type-checking, but its information that the programmer is up to supply, and tends to be the part of the optimization stage rather than normal, incremental development. Lisp is usually defined by its dynamic nature, and run-time type checking is a big part of this. Making the programmer have to worry about how objects are managed before the fact would conflict with the freedom and expressibility one would expect from such a system. It also makes the compilers themselves simpler in comparison: Normal code without any extra declarations is safe by default, some systems might ignore such declarations for always being 'safe' (say if you're on a bytecode VM like CLISP or a Lisp machine that does hardware type-checking). SBCL compiles code quite quickly--so I've heard others tend to be even faster; the Rust compiler on the other hand is more likely to introduce a young programmer to the concept of thrashing. I really see them as two mutually incompatible worlds although they may not seem to be at first glance. One thing to remember is that Lisp is essentially the flagship "The Right Thing" language, while C is the "Worse is Better" language. Rust is neither, it is something entirely different which I think is overdue for a name, perhaps something that reflects the losing characteristics of both philosophies.

(This isn't to discredit the OP article: it's still a cool hack!)

> perhaps something that reflects the losing characteristics of both philosophies.

Oof. With how much love Rust gets here, I didn't expect to see it being called out like that.

How about "The Worse Thing"?

> Some Lisp compilers like SBCL are already capable of more extensive compile-time type-checking, but its information that the programmer is up to supply

Which is nice and all, but very much gimped by the glaring holes in CL's typing tooling: you can't create actual types, only derived types (deftype) and struct/class types. The two consequences of that is that you can't type cons-based lists/trees (arguably THE Lisp data structure) because deftype can't be recursive and you can't create parametric types, it's an internal thing only used for https://www.lispworks.com/documentation/HyperSpec/Body/t_arr... (and not even hash-tables, these are completely untyped!).

> you can't type cons-based lists/trees (arguably THE Lisp data structure) because deftype can't be recursive and you can't create parametric types

  (deftype list-of (type)
    `(cons ,type (or null (list-of ,type))))

  (typep '(1 2 3 4) '(list-of integer))
  => T

Most of the standard types from which derived types can be made are parametric types, which specialize on their arguments in different ways. They work the same way as macros. One upon a time I wrote a type specifier that allows you to specialize on a lambda expression, using the ordinary 'satisfies' type that only lets you provide named functions:

  (deftype satisfiesp (function)
    (let (predicate)
      (cond ((symbolp function)
             (setq predicate function))
            ((functionp function)
             (setq predicate (gensym))
             (setf (symbol-function predicate) function))
            ((and (consp function) (eq (car function) 'lambda))
             (setq predicate (gensym))
             (compile predicate function))
            (t (error "~S is neither a lambda expression, function or symbol." function)))
      `(satisfies ,predicate)))

Now you can even type-check on lambda expressions, and quasiquotation can sidestep the issue of capturing variables.

  (defvar foo 'bar)
  (typep 'bar `(satisfiesp (lambda (x) (eq x ',foo))))
  => T

https://www.cs.cmu.edu/Groups/AI/html/cltl/clm/node44.html

The CLHS sez "Recursive expansion of the type specifier returned as the expansion must terminate, including the expansion of type specifiers which are nested within the expansion." though. As expected, here on SBCL, your typep blows the stack.

> Most of the standard types from which derived types can be made are parametric types

I meant "parametric with a type parameter", but yeah.

> satisfies

I was (obviously, I hope) talking about static, not runtime typing.

Rust is definitely not a simple language, I agree there.

I am quite likely a bit naïve here, but I'm having trouble understanding why you hate that this is possible. Now, the macro system definitely can generate near infinitely-complex code, but I'm not getting how implementing a sandboxed lisp using macros is a particularly potent example of the language being less manageable than at its genesis.

On another note, the fact that the type system is turing-complete, like with C++ templates or the Haskell type system (variously dependent on which GHC languages extensions you're using...), makes me want to see a lisp implemented that way!

Implementing a lisp in the type system would be fun, that's originally what this project was about until I got distracted with macros. Awesome projects like fortraith (https://github.com/Ashymad/fortraith) already exist, and even far more useful projects like dimensioned (compile time dimensional analysis using type level numbers) are inspiring. These examples, although cool, are probably a worse sign for krick for Rust compared with macro_rules being Turing complete.

Aha, that's pretty cool! Didn't even scroll more halfway down the page before my lizard brain saw some something resembling Peano numbers.

Thanks for sharing your project, by the way - between this and the other one you linked, I think I know what my leisure time this weekend is going to consist of...

Oh, I'm very well acquainted with Sectorlisp - I even have a scrappy local package on nixos to build, boot and launch a qemu repl for it locally! Somewhere around here there's an HDD in an X200 that probably still has it in the boot sector on its HDD.

I think I objectively suck at lisp, but that doesn't preclude me from being a "Greenspan enjoyer" :) I think all of my projects that used Selenium ended up with a haphazardly-pieced-together lisp interpreter as their scripting tool after I realized whatever I was writing was effectively becoming a DSL in a .ini/.yaml/.toml file. These days, I mostly use nix, but my dream is a lispy NixOS (yes, I know of and have used Guix, but I really just want a lisp<>Nix 'native' compiler, and to be able to use it with nixpkgs without hassle).

> but I think it started as something way more manageable than what it has become.

Hard disagree here. Rust team is continuously making the language easier to work with by removing limitations and making features more orthogonal. A notable example is non lexical lifetimes, impl Trait in return position or async traits. Also they have removed some features eg before it went 1.0, it even had GCed references built in with special syntax.

The only real change since 1.0 was async. If you want to live without async that is completely up to you. It is an entirely optional part of the language.

If you want a language guided by the principle of simplicity Rust was never that and you have plenty of options elsewhere.

To be fair, neither are C preprocessor macros. They're instead token substitutions. It's not that much better, but they're not literally text substitution. They're at least more clever than just that.

They're also of course surprisingly powerful, at the expense of being very cludgy.