Style: indentation of multiple-arity functions

[wjlroe]

This pull request adds guidelines on how to indent multiple-artity functions. The exact style here is to vertically align arguments and function body - as elements of any sequence would be indented; however this is up for discussion as there is no de facto agreement on how to indent these forms. Different tools have arbitrarily chosen different indentation styles and it would be good to reach an agreement on what most people prefer so that we can all conform to a consistent code style.

[amoe]

This indentation style seems logical and attractive, IMO it should be merged. 👍

[Gonzih]

I think default style of using 2 spaces is much more intuitive in this case.
Also is simpler and probably supported by most of the tools by default.

[bbatsov]

I think default style of using 2 spaces is much more intuitive in this case.

Why is it intuitive? It differs from the standard function structure. People have the tendency to confuse things that they're used to with things that are actually intuitive.

Also is simpler and probably supported by most of the tools by default.

A discussion in clojure-mode's issue tracker makes me believe that no tool except clojure-mode indented code this way (and clojure-mode doesn't do this anymore as well).

[Gonzih]

Ok, my bad about tools support.
If it's only specific to clojure mode then just merge it.

On 12/03/2014 03:38 PM, Bozhidar Batsov wrote:

I think default style of using 2 spaces is much more intuitive in
this case.

Why is it intuitive? It differs from the standard function structure.
People have the tendency to confuse things that they're used to with
things that are actually intuitive.

Also is simpler and probably supported by most of the tools by
default.

A discussion in clojure-mode's issue tracker makes me believe that no
tool except |clojure-mode| indented code this way (and |clojure-mode|
doesn't do this anymore as well).

—
Reply to this email directly or view it on GitHub
#95 (comment).

[zane]

In the absence of any clear reasons not to, matching the indentation style for single-arity functions seems like the right way to go here.

[bbatsov]

of a function definition
arguments -> parameters

[bbatsov]

I agree with the PR in general, but I've added a bit of stylistic comments.

[wjlroe]

@bbatsov I can rebase those commits if you'd like

[bbatsov]

many -> two

[bbatsov]

Ping :-)

[wjlroe]

Whoops - fixed

[bbatsov]

Yes, please.

[wjlroe]

Rebased

[bbatsov]

P.S. A better commit message would be "Add a rule about the indentation of multi-arity functions"

[wjlroe]

Sure, happy to change it. I did a quick browse of the git history to see if I could divine a pattern

[bbatsov]

👍

[shaunlebron]

This isn't representative of what people are doing in the core libraries. They indent the body after arguments in multi-arity functions. It's everywhere in clojure, clojurescript, core.async, tools.reader, and probably more.

In fact, clojure.pprint will indent one space after the arguments in multi-arity. Paste this code into the online clojure.pprint service to see it:

(defn symbol
  ([name]
   (if (symbol? name)
     name
     (let [idx (.indexOf name "/")]
       (if (== idx -1)
         (symbol nil name)
         (symbol (.substring name 0 idx)
                 (.substring name (inc idx) (. name -length)))))))
  ([ns name]
   (let [sym-str (if-not (nil? ns)
                   (str ns "/" name)
                   name)]
     (Symbol. ns name sym-str nil nil))))

after clojure.pprint:

(defn symbol
  ([name]
    (if (symbol? name)
      name
      (let [idx (.indexOf name "/")]
        (if (== idx -1)
          (symbol nil name)
          (symbol
            (.substring name 0 idx)
            (.substring name (inc idx) (. name -length)))))))
  ([ns name]
    (let [sym-str (if-not (nil? ns) (str ns "/" name) name)]
      (Symbol. ns name sym-str nil nil))))

[bbatsov]

Much code in the wild was affected by an unfortunate bug in clojure-mode. Who knows, maybe even pprint was modelled after it. I'm one hundred percent certain that the current indentation is the most reasonable (and intuitive) way to indent such code, as it's consistent with the way single-arity functions are indented.

[marick]

I find the new style from clojure-mode makes multi-arity functions harder to read because the different arglist-body pairs don't stand out as well (especially for shorter ones).