Note to self: learn a lisp. Lisp people appear to be having more fun than the rest of us.
If you're serious about learning a Lisp but motivated more by, as you say, "having more fun" rather than, say, landing a six-figure job writing it professionally... then may I recommend Janet[0] for your consideration. Janet is an embeddably-small, yet surprisingly batteries-included Lisp implemented in pure C. In terms of syntax and core library it borrows more directly from Clojure than from Scheme, but all the modern Lisps have their bits of influence. I've found both the language and the tiny little community that exists around it delightful.
As an example of the latter, somebody smart wrote a real actual book[1] about Janet recently that was on the HN front page for a day or so when he first released it. It's a gentle introduction not just to Janet but to Lisp in general, and assumes only general proficiency with JavaScript to get you up to speed. I recommend it.
Because it's not built around linked lists, the core type most encountered for lists of data is arrays and tuples. Neither is conducive to efficiently removing elements mid-set, or composing two sets, or interleaving two sets, or other operations that require reordering, replacing, adding or removing elements in the set. When I'm writing code in lisp I don't think about that overhead much, because for linked lists it's not an issue.
> Insert a sequence of random integers into a sorted sequence, then remove those elements one by one as determined by a random sequece of positions: Do you use a vector (a contiguously allocated sequence of elements) or a linked list?
The key is that the sequence is sorted; and is kept sorted throughout the insertion process.
In the contrived example, a linked list is clearly a terrible option for manipulating a sorted set of integers. Modern computers can slice and dice contiguous integers with vectorized routines, and those benefits are lost with linked lists that are storing their data haphazardly across the heap.
If your problem space is best defined with contiguously-stored numbers, then for sure, don't use linked lists. Most lisps will happily provide you with vectors and arrays for these use cases.
And yet a subset of the linked list called a tree is usually the right answer to problems.
Janet's REPL has a debug mode, but I'm sadly not qualified to evaluate whether/how it measures up to CL's due to my own inexperience with either. :)
As for restarts (I had to Google around to get an idea of what that means)—it seems to me that Janet does not have first-class support for restarts in the way that some other Lisps, for e.g. CL, do. Presumably (as it would be in most other languages I would suppose) one could recreate that experience in Janet by tapping into the first-party error handling and REPL primitives. But you'd definitely be rolling your own rather than having it already in Janet out of the box.
What Lisp should we learn if we want to make money from it?
That said, I don't think it matters much. A developer familiar with some lispy language (and perhaps functional programming) should be able to quickly pickup any other lispy language. And the developers that make the most money have probably used a lot of programming languages, with different paradigms.
[0]: https://survey.stackoverflow.co/2022/#top-paying-technologie...
[1]: https://survey.stackoverflow.co/2023/#section-top-paying-tec...
I got my first job furiously doing that for nights on end. It felt like cheating but it’s ultimately way faster than waiting potentially years to figure out those tricks on your own.
Edit: like a hash-collision for geeks
> The goal of the Make-A-Lisp project is to make it easy to write your own Lisp interpreter without sacrificing those many "Aha!" moments that come from ascending the McCarthy mountain. When you reach the peak of this particular mountain, you will have an interpreter for the mal Lisp language that is powerful enough to be self-hosting, meaning it will be able to run a mal interpreter written in mal itself.
~ https://github.com/kanaka/mal/blob/master/process/guide.md
Interacting only with something you have made is the worst possible way to learn anything about existing, mainstream Lisp.
Not knowing anything about the prior art will practically ensure that your own thing is a collection of quirks specific to it, and that then ensures that your Lisp knowledge is entirely rooted in and limited to that thing.
You're just not going to single-handedly reinvent things that took numerous hackers over several generations to figure out. You may be "good", but so were they, and there is only one of you.
You will not even be able to converse with actual Lisp people, due to not having the right concepts and terms.
I feel MAL does a decent job illustrating the core concepts of Lisp, but then I'm not a Lisp expert by any means.
I don't have the impression that the author of the MAL project used it as a way of learning Lisp.
To my best understanding, MAL provides a step by step recipe and test cases for implementing a language dialect, plus example implementations for which that has been done.
Someone going through the MAL exercise will mainly interact with the chosen implementation language. Writing a new MAL implementation could be used as an exercise to learn some unfamiliar language.
I'm not convinced that programmers implementing MAL are actually learning how to use MAL, since it doesn't look as if the exercise requires them to write MAL code to solve problems. To learn MAL, you would take the project as-is with its integrated implementations and use that to get other work done.
Suppose someone actually learns MAL by working with it. Are they learning Lisp?
I'm not sure how much of their acquired skill will transfer to working with a mainstream Lisp. MAL looks rather like a kind of Mock Lisp (see: https://en.wikipedia.org/wiki/Mocklisp). It's implemented even in Bash and Awk using string processing.
The good thing about MAL is that at least the participants in the exercise are following a specification instead of just making something up as they go along and calling it Lisp. Making something that is compatible with a spec and other implementations is a good exercise for software engineering students, regardless of what it is. MAL would still be valuable that way if it was, say, MAVE: make a vi editor, or MAM: make a Mario game.
I have no idea how effective MAL is educationally in teaching the concepts themselves that underpin MAL, which revolves around this question: is it possible to mechanically follow the recipe and get a new implementation working, without understanding the concepts? Separately from the question of whether MAL concepts are Lisp concepts, is the MAL implementor who follows the structured workflow learning the MAL concepts, or are they just massaging code to get some tests to pass. (But, of course, even if they are, so what; nothing stands in their way of learning any concepts they want in any other manner.)