“JavaScript Object Notation” is widely used. It is used as a data interchange standard and is also human-readable. One thing I dislike about it, is the way it’s usually formatted.

jq

jq is the most popular tool to format and query JSON.

When formatting, it puts commas at the end of a line. Brackets ([,]) and curly braces ({,}) are on their own lines (line 1) or together with a field key (line 4).

jq formatting
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12

{
  "credits": {
    "director": "Steven Spielberg",
    "starring": [
      "Sam Neill",
      "Laura Dern",
      "Jeff Goldblum"
    ]
  },
  "released": 1993,
  "title": "Jurassic Park"
}

my Preference

I personally prefer the formatting to be similar to Haskell lists. The opening bracket ([) / curly brace ({) is on the same line as the first element of the starting object / array. Leading Commas are indented the same as the opening bracket / curly brace.

my preferred formatting
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11

{ "credits":
  { "director": "Steven Spielberg"
  , "starring":
    [ "Sam Neill"
    , "Laura Dern"
    , "Jeff Goldblum"
    ]
  }
, "released": 1993
, "title": "Jurassic Park"
}

Why do I like this better?

  • The structural components (object, array) can always be recognized on the left, while the content (string, number, boolean) is always on the right.

  • Single line JSON values are on the same line as their key.

  • Multi line JSON values are on their own lines. Compare line 4-8 by jq and line 4-7 by me.

  • Line based diffs are slightly more efficient.

  • There are shorter and fewer lines overall.

  • Trailing commas are easily forgotten.

Last note

I can totally understand, if you call this article nit-picking. It’s definitely more important to agree on one kind of formatting, than to use the optimal kind of formatting. It’s just a minor thing that has bugged me before.