• mrh@mander.xyz
    link
    fedilink
    English
    arrow-up
    2
    ·
    6 months ago

    Common Lisp “solves” most language-level problems by providing metaprogramming capabilities via lisp-style macros. (Almost) any language feature you would want can be implemented with lisp macros, and many such features already have been. So you don’t have to worry whether or not lisp has “for i in…” loops, or pattern matching, or generics, or virtually anything else, because if it doesn’t, you can write it! Plus if it’s really a good feature somebody has probably already made a library for it (if it’s not already part of the standard).

    One of the most extensive examples of this is Coalton, which is an ML-style statically typed EDSL for Common Lisp.

    There are metaprogramming features in a few other languages: template haskell, C pre-processors, even macros in Rust or Julia. But these all fall very short of lisp-style macros because those languages are not (truly) homoiconic, which makes the macros awkward to write and integrate into the language. This kind of metaprogramming is rarely employed, and when it is only for heavy duty tasks, and even then is generally discouraged as a last resort/special circumstance. But lisp macros are very easy to write because it’s the same as writing any other piece of lisp code. This results in macros being used often for smaller lightweight abstractions in the same way you write a small function.

    The other big pro of lisp is image based development. But that’s not so much solving a problem in other languages as it is simply a feature that they don’t (and pretty much can’t) have.

    And all of this is done in a language with less syntactic and semantic primitives than almost any other language, including the other “simple” ones like Python, Ruby, Elixir, etc.