Comparing Arrays in JSON Diff

Arrays are the most challenging data structure for JSON diff algorithms because, unlike objects where keys provide natural identifiers, array elements are identified only by their position (index). Different diff strategies produce significantly different results.

Index-based comparison (naive):

The simplest approach compares elements at the same index. If the array grew or shrank, trailing elements are reported as additions or removals:

// Before
["apple", "banana", "cherry"]

// After
["apple", "blueberry", "banana", "cherry"]

Index-based diff reports:

• Index 1: changed "banana" to "blueberry"
• Index 2: changed "cherry" to "banana"
• Index 3: added "cherry"

This is technically correct but misleading, as the actual change was inserting "blueberry" at index 1.

LCS-based comparison (smart):

More sophisticated algorithms use the Longest Common Subsequence (LCS) algorithm (similar to what git diff uses for lines of text) to find the optimal set of edit operations. LCS correctly identifies:

• Index 1: inserted "blueberry"

Identity-based comparison (objects in arrays):

When arrays contain objects, many diff tools let you specify an identity key (like id) to match elements across the two arrays regardless of position:

// Before
[
  { "id": 1, "name": "Alice" },
  { "id": 2, "name": "Bob" }
]

// After
[
  { "id": 2, "name": "Robert" },
  { "id": 1, "name": "Alice" }
]

With identity-based diffing on the id field, the algorithm correctly reports that Bob's name changed to Robert and the order swapped, rather than reporting both elements as completely replaced.

Best practices:

• Use identity-based diffing for arrays of objects whenever possible.
• Be aware that large arrays with no identity keys will produce noisy diffs.
• Consider sorting arrays before comparison if order does not matter to reduce false positives.