flavio81 11 months ago [-]
The author, a famous and well-liked lisper, is not considering portability features. CL is an ANSI standard and code often runs with no changes in many distinct CL implementations/compilers/interpreters.
Also, related to that point: There are many different CL implementations out there that satisfy different use cases, like for example JVM deployment (ABCL), embedded systems (ECL), speed(SBCL), fast compile times (Clozure), pro-level support (LispWorks, ACL), etc. So the same code has a huge amount of options for deployment. It really makes Common Lisp be “write once, run anywhere”.
Then speed is also never mentioned. Lisp can be seriously fast; under SBCL it is generally 0.3x to 1x C speed; LispWorks might be faster, and there’s a PDF out there called “How to make Lisp go faster than C”, so that should give an idea of Lisp speed potential.
CL wasn’t created by philosophing about what a programming language should be for many years; CL was basically created by merging Lisps that were already proven in the industry (Maclisp, Zetalisp, etc), already proven to be good for AI, heavy computation, symbolic computation, launching rockets, writing full operating systems, etc.
CL is a “you want it, you got it” programming language. You want to circumvent the garbage collector? Need to use GOTOs for a particular function? Want to produce better assembly out? Need side effects? Multiple inheritance? Want to write an OS? CL will deliver the goods.
In short, I would guess that from a computer scientist or reseacher point of view, Racket is certainly more attactive, but for the engineer or start-up owner that wants to have killer production systems done in short time, or to create really complex, innovative systems that can be deployed to the real world, Common Lisp ought to be the weapon of choice!
— Why I haven’t jumped ship from Common Lisp to Racket just yet
— Hacker News
2022.07.13 Wednesday ACHK