JSON Diff

Structural comparison, not text comparison

Running a text diff on two JSON files reports every reformatting as a change. Object key order carries no meaning in JSON, so two documents with identical data but different key order are equal, and a line-based diff will insist they are not. A structural diff parses both sides and compares the resulting values, which is the only comparison that answers the question you actually asked.

Arrays are ordered, objects are not

This is the distinction that governs everything. Object keys can appear in any order and the documents are still identical. Array elements cannot: moving an element changes the document. So a diff must treat reordering inside an object as no change and reordering inside an array as a real one, which is why comparing arrays of objects is the hardest case and where tools disagree.

Types matter as much as values

The string "1" and the number 1 are different values, and a diff should say so. This catches a genuine and common class of API bug, where a field silently changes type between versions and everything downstream keeps working until something tries arithmetic on it. Similarly, null, an empty string and a missing key are three different states that get conflated constantly.

Numeric precision

JSON numbers parse as doubles, so two documents can differ in a large integer that both display identically. Any value beyond 2^53-1 is at risk. If your payloads carry 64-bit identifiers, compare them as strings or the diff will report equality where the raw bytes differ.

Compared locally

Both documents are parsed in your browser and nothing is transmitted. API payloads routinely contain tokens and personal data, so sending two of them to a remote service to find a difference is an unnecessary disclosure.