This is the way Clojure works too. Nil is treated as an identity/neutral element. This goes beyond lists: in contrast to Common Lisp who equates nil to the empty list, Clojure maintains a distinction and you're supposed to handle nils by correctly implementing functions. This is as easy as using clojure's core functions, they all have been tailored to handle nil as the neutral element.

As a consequence, in Clojure, nil's neutrality extends to other datastructures, for instance (hash)maps. (assoc nil :some :value) => {:some :value}.

I like the idea, but JVM is a show stopper for me :/

May I ask why? I've no dog in this fight (I don't work with Java) but I've seen plenty of people against Java the language who were very happy to use Scala or Clojure. I haven't encountered anyone who has specifically said they were against the JVM, so I'm interested whether it's something technical (maybe you do more real-time or work with resource-constrained systems) or if it's something more personal. There are no bad answers, I'm not looking for an argument just genuinely curious :)

Copy pasted: Why => because JVM is poor no fit for CLI, and is a poor fit for games. Those two fields in which I see myself using common-lisp.

Unsure what's so wierd about it, everyone (most) knows about slow cold startup times, and inability to control memory layout in JVM. Also, horrible C bindings

I didn't say it was weird :(

Just was curious if it was one of the commonly encountered things (which you listed) or if there were some other limitations I wasn't aware of.

Why? I have no particular feelings for or against the JVM. It always surprises me when people get so passionate about runtimes - how much do you interact with them? They're just in the background doing their thing.

I can see reasons for not wanting to use the JVM in specific cases. For example, it may not have the performance you want for the application you're developing.

I'm just confounded by general statements such as "JVM = no go", "I want to use C# but I'm more of a JVM guy than dotnet" (or vice versa).

Is this some bizarre nu-console war that's going on in the programming community that I have missed?

Why => because JVM is poor no fit for CLI, and is a poor fit for games. Those two fields in which I see myself using common-lisp.

Unsure what's so wierd about it, everyone (most) knows about slow cold startup times, and inability to control memory layout in JVM. Also, horrible C bindings

For me it's microcontrollers, however I've heard in spite of the unmaintained status, clojure-scheme[1] is very usable. It uses gambit to produce C-code you can compile anywhere.

https://github.com/takeoutweight/clojure-scheme

I'm with you on stock JVM. The slow startup time is often a deal breaker for me. But there are projects that address this. Like https://github.com/babashka/babashka for Clojure. Or GraalVM for general JVM use.

How do you control memory layout in Common Lisp?

you might be interested in embedded Schemes. I use s7, which shares a lot of features with both CL and Clojure (for a Scheme) and is very easy to embed in C or C++ apps. I use it for music coding, but others (e.g. Naughty Dog) have done it for games. It's a super fun way to develop. My high level code is mostly Scheme, interfacing with low level C.

I don't see how the JVM is a "poor fit for games" when it's entirely contextual towards the goals of the game itself. Garbage collection also does not automatically disqualify an environment from games development, and there are plenty of games made in garbage-collected languages out there by both hobbyists and professionals using frameworks like LibGDX and Monogame/FNA/XNA. Depending on the kind of game, sure, the JVM might not be the right choice (or a choice at all), but for most people they likely will be too busy trying to actually finish the game instead of tuning for performance.

> Why? I have no particular feelings for or against the JVM. It always surprises me when people get so passionate about runtimes - how much do you interact with them? They're just in the background doing their thing.

I spent a day writing Clojure after watching a bunch of talks about it. Very keen on persistent data structures.

Every time I put a foot wrong it printed a Java stack trace for me, which was very large but told me nothing. I then went looking for the optimiser and discovered there wasn't one, it just falls through into the JVM JIT and hopes for the best. And finally it turned out a bunch of Clojure libraries were reskinned Java, and therefore likely to have the stereotypical properties of Java code. So that all seems directly attributable to the runtime.

Clojure the language is totally my sort of thing but that was such a horrendous experience that I haven't picked it up again in... what turns out to be seven years (found my stack overflow question about it).

It's littered, just hidden. Eg: if, if-let, when, when-let, unless etc. This makes it appear that the data is passed down in a monadic style, but it often isn't; *nil* is not an actionable value.

I don't usually find myself needing a null element in Lisp, so there's no need to handle any. (Ditto Java, but I don't write much Java to be fair.)

Common Lisp code does have checks for NIL. When they are missing you get bugs like this one:

https://bugs.launchpad.net/sbcl/+bug/2002244

https://sourceforge.net/p/sbcl/sbcl/ci/06f1639e53e99365f48a1...

If you are calling a generic function you can sometimes avoid such checks by introducing methods specialized on the type NULL (which contains only NIL).

Agreed. After 16 years of programming experience, I've somehow settled into a world where error handling is no big deal in my software:

- Let-it-crash philosophy of Erlang/Elixir

- Monadic errors of Rust

- Conditions/restarts/REPL model of Common Lisp

Using Go (or C or Python or anything with exceptions) feels like going back to the 1970s.