JSON Diff Tool - Online JSON Data Comparison

1. Compare two JSONs

1. Old config on the left, new on the right; differences list live by field path.
2. Reading: added means the right side has an extra field, removed means the left side had one, and a path like config.timeout with "old → new" is a modification.
3. Syntax errors on either side are reported separately — both must be valid JSON for the comparison to mean anything.

2. How it splits work with text diff

• Structural comparison ignores key order and indentation: {"a":1,"b":2} versus the same data pretty-printed is "no difference" — exactly where it beats text diff.
• Conversely, when the format itself matters (indent style, key order), use text diff; when the data matters, use this tool.
• Arrays compare positionally by default: reordered elements show as modifications; for order-insignificant arrays (tags, permission lists) enable "Ignore array order" to match as multisets.

3. Noise-free diffing and JSON Patch export

• Ignore keys: list churn fields like updatedAt, traceId — skipped at any depth, leaving only signal in the diff list.
• Array key: for shuffled object arrays, align on a business key (e.g. id) — same-key rows report field-level diffs like users[id=7].name instead of whole-row add/remove.
• Copy JSON Patch: one click exports an RFC 6902 operation sequence ready for fast-json-patch replay; note the patch is unavailable in "ignore array order" mode since paths carry no index semantics.
• Leftover-element pairing in ignore-order mode is an O(n²) greedy pass — pair it with an array key for shuffled arrays of a thousand elements or more.

4. Choosing among four diff granularities

JSON comparison has four levels: textual (line-diff the raw strings) for formatting changes; key-level (compare by key path) for response regression; semantic (numeric tolerance, arrays as sets) for floats and unordered arrays; structural (schema shape) for add/remove audits. This tool aligns at key level — the mainstream need during integration debugging: which field changed between two versions of a response. Array alignment is the biggest variable in key-level diffing: index-by-index (strict but order-sensitive) versus aligning on a business key (say id) can yield entirely different results — confirm whether the array ordering is guaranteed before comparing.

5. The standard integration-debugging loop

1. Save the same endpoint response from staging and production as two JSON files.
2. Diff them here and jump straight to the first divergent path — two orders of magnitude faster than eyeballing thousands of lines.
3. Trace each divergent field: dirty data or a serialization-logic change.
4. After the fix, replay the diff and confirm the delta collapses to zero.

6. Option combos and acceptance

Combine ignore-keys, array align-key, and ignore-order carefully: drop churn fields first, align object arrays on a business key, then treat unordered tag lists as multisets. JSON Patch needs stable indexes — unavailable in ignore-order mode. Validate both sides as JSON before reading the list; replay until paths collapse.