Dunno if I'm in the minority, but I just can't warm to [], either in lieu of or augmenting, () in SEXPRs. Just never felt they were necessary and, if anything, I find them distracting. Likely because I'm not used to them (baby duck syndrome and all), but the regularity of Lisp syntax, to me, is a boon, and the [] just mess that up for me.
Reader macros/literals allow for explicitly differentiating between built-in data types, and provide semantic meaning to otherwise indistinguishable forms. For example, defining a function:
In Racket, you write `(define (foo arg1 arg2) (+ arg1 arg2))`. Why isn’t `(foo arg1 arg2)` evaluated as a function? It’s a list and it isn’t quoted, so it should be evaluated. However, `define` is a macro (I think, might be a special form), so that list isn’t evaluated and doesn’t have to be quoted. That violates the basic tenant of “unquoted forms wrapped in a list are evaluated as fiction calls.”
In Clojure, you write `(defn foo [arg1 arg1] (+ arg1 arg2))`. The [] is a reader macro, a vector literal. It communicates that it’s not a function call but simply data (in this case, a vector containing two symbols).
Maybe I’m just used to it but I find this stuff really helpful.
I would agree more with this view if there was also an explicit indicator for lazyness: in (define (foo arg1 arg2) (+ arg1 arg2)) the two parts (foo arg1 arg2) and (+ arg1 arg2) are both interpreted specially; the former as a data structure and the latter as code with delayed execution.
Using square brackets in place of some parens is something some of the Racket professors started doing. I assumed it was because they thought it would be helpful to teach high school students in http://htdp.org/ . I defer to them on teaching, but I think it looks ugly, is harder to type, and seems like it's more confusing in code examples (people thinking it's significant, when it's not).
One thing I did do is make an Emacs mode, https://www.neilvandyke.org/quack/ , in which typing square brackets instead inserts parentheses when appropriate. This is helpful on US keyboards, on which parens require a multi-key press but square brackets do not.
Yeah, in R6RS square brackets are considered to be perfectly interchangeable with regular parentheses. That decision was reversed for R7RS-small, though.
I find them very useful in the way Racket uses them, which is by convention only. In Racket, `[<expression>]` is identical to `(<expression>)`. But Racket uses them in a conventional sense, for example, in `let` bindings:
