Yup, but when you have to integrate with things you can't control, the advantages are fewer, especially compared to something purpose-built for that thing and t

Yup, but when you have to integrate with things you can't control, the advantages are fewer, especially compared to something purpose-built for that thing and the trade-offs can easily turn away from Lisp. e.g. you're going to have a much easier time making a game in plain Godot than you would trying to mix CL with Godot's native API, or if you're crazy doing an entire 3D engine + game in CL. Even in that last approach, integrating with the graphics stack on modern systems is going to be nuts. Going further into OS-facing stuff like sound, rendering text, handling multiple languages, handling inputs from mice/keyboards/controllers, dealing with multiple monitors, microphones... it's a good thing Common Lisp has a decent FFI (it's even possible, though more difficult, to integrate with pure C++ stuff using SBCL and not having to switch to Clasp, but it's nice that Clasp exists in the same way it's nice ABCL exists to more easily integrate with Java stuff) because you will be using native libraries for some things, and it will suck to some degree because those are written in the world and mindset of dead, static programs. But the pure Lisp side will be great, and you'd be using those libraries or something like them anyway if you switched Lisp out. An underrated Hickey talk is his Effective Programs one: https://github.com/matthiasn/talk-transcripts/blob/master/Hickey_Rich/EffectivePrograms.md https://github.com/matthiasn/talk-transcripts/blob/master/Hi... It's another entry in my personal (and useless time-wasting) fascination with wondering what could have been if CL won in the 90s. No need for Java to come around and "get C++ programmers half-way to CL", with excruciatingly slow followups year after year that finally by Java 8 and after got us something (when also using additional things like a giant IDE with good refactoring tools and simple code generation (no manually writing getters/setters or managing a giant list of imports), Spring, JRebel, and sometimes AOP frameworks) actually not too unpleasing compared to a CL experience. But most things were always there in the CL experience in the 90s, especially if you could shell out for the commercial systems, it really is just Java playing catch-up. And then you get in that talk and elsewhere a story about how non-technical decision making didn't even give Lisp systems a chance in production, so as a last resort Lisp was put on top of the JVM which was accepted. Clojure is a great achievement, and ABCL came too late, but it's kind of a shame either has to exist. Still, the thousands upon thousands of man-years have been spent, and because of that Lisp's advantages are fewer. However the Lisp world itself hasn't been static since the 90s either, and so the counter-advantages of vastly larger ecosystems have grown fewer as well. For example, it took until as late as 2010, but finally quicklisp was made and the problem of libraries and finding/getting them was dramatically lessened. For comparison, Perl's CPAN is from the 90s, Python's PyPI is from 2003, Maven 2004... Lisp was late. But the last 10 years of Lisp seem to be going a lot better than the 10 before, and while there are still problems, they don't really look like uniquely Lisp problems, and supposedly unique ones like supposedly not being able to do infix arithmetic are addressed with slow education to the contrary (while I'm here might as well link for the nth time, the solution available since 1993: https://github.com/quil-lang/cmu-infix https://github.com/quil-lang/cmu-infix). It was such education years ago from a HN comment that made me give CL another look and then never look back at Clojure/Scheme/etc. (though Java is still my preferred breadwinner should I find myself suddenly needing more income). (The comment was simple, just showcasing that yes in fact CL has typing support, the compiler is available at runtime, and with SBCL the compiler can report not just about wrong types / typos but dead code elimination and failures to apply optimizations. Cool stuff. Made me realize I never knew CL at all. Even CLOS, the OOP system, had some pleasant surprises when I finally got around to learning it and seeing it's far more than what other langs offer.) Meanwhile the advantages are still there.