On: The Racket Manifesto

[stcredzero]

The problem with Lisp-like DSL construction, is that the interface doesn't play well with many populations of domain experts. Even Smalltalk, with its syntax designed to be usable by children can fulfill the technical side of constructing DSLs easily, but still leaves something to be desired when it comes to domain expert use. (Various Smalltalks had their rough equivalent to #lang.)

(I know there is at least one Ruby based DSL which made some inroads to mobile device software testing.)

Basically, DSLs tend to only get used by experts who are already programmers or otherwise technically disposed. Maybe we're getting to the point where rudimentary coding is just going to be a new kind of literacy?

[shriramkmurthi → stcredzero]

This is a criticism of Lisp-based DSLs but not of Racket-based DSLs. Racket actively enables you to impose different, non-parenthetical surface syntaxes on the DSL. It is not a "this sort of works if you arrange it carefully _and nobody screws anything up_" thing like in Ruby, it really is an explicit language definition.

The easiest way to understand this in action is to read about Danny Yoo's Brainfuck embedding in Racket: https://www.hashcollision.org/brainfudge/

[muraiki → shriramkmurthi]

I really liked the following article, which creates a text adventure language in Racket that looks like this:

  ===PLACES===
  ---desert---
  "You're  in a desert."
   [cactus, key]
  north    start
  south    desert

http://queue.acm.org/detail.cfm?id=2068896

[shriramkmurthi → muraiki]

Scribble is another great example. It's amazing in its own right, and by being embedded into Racket, gets all the Racket benefits: for instance, separate compilation! (Thirty years and waiting, LaTeX.)