Hacker News Reader

Comment by catch23 | Hacker News Reader

I think languages that allow the use of non-alphanumeric characters as method definitions make the language appear more magical simply because it looks like built-in language syntax. I remember when I first looked at sbt, they were creating directory paths with the '/' as an operator to concatenate strings. (ie "home" / "user") I'm guessing that might also be the reason Ruby is cited as magical all the time.

View on HN

I really can't stand this trend. When I took a look at Lift, it was riddled with these "magical" method definitions that look nice but mean absolutely nothing to someone who's new to the framework. I'll gladly take a self-explanatory alphanumeric method name over a "/", a "+", etc.

Why not have both? Also do you happen to use any regular expressions ever? Can you imagine specifying a regular expression using self-explanatory alphanumeric method names?

Probably true, yes. For instance, the apply-transformation

  x.apply(y) == x(y)

and the way operators are handled

  1 + 2 == 1.+(2)

are fairly powerful syntactical tools which allow seemingly magic DSL stuff. Many enter programming via Java and C; languages that are far less malleable and open to seemingly built-in modification than, say, Lisp, Scala or Ruby. However, flip side of syntactical magic? Stuff like '·' (which I recently encountered in some Scala library).

Or Databinder Dispatch:

  Http(:/("google.com") / "bla" / "bar" << "why=so&x=y" ># foo)

Nice library, but, you know, conciseness isn't everything. (By the way, this places a POST request to http://google.com/bla/bar with request body "why=so&x=y" and delivers the response as a parsed JSON tree to foo.)

Reply on news.ycombinator.com