Test ordered_json binary formats, flatten, and patch

[22elix3r]

Summary

unit-ordered_json.cpp only checked insertion order and one compile regression. Binary formats, flatten/unflatten, patch, and update(merge_objects=true) were exercised almost exclusively with default nlohmann::json, so ordered_map-specific behavior could regress silently.

Related Issue

Fixes #5421

Changes Made

• CBOR, MessagePack, BSON, and UBJSON round-trips that assert insertion order is preserved
• flatten/unflatten key-order checks
• diff/patch and patch_inplace round-trips
• update(..., true) recursive object merge on ordered_json

Testing

Commands executed:

• cmake -S . -B build -DJSON_BuildTests=ON -DCMAKE_BUILD_TYPE=Debug
• cmake --build build --target test-ordered_json_cpp11
• ./tests/test-ordered_json_cpp11 --no-skip -tce='*downloaded*'

Results:

• 4 test cases passed, 31 assertions, 0 failed

Notes

This does not add every binary-format edge case listed on the issue (duplicate keys in binary maps, BJData, std::format). It pins the order-sensitive APIs most likely to differ from std::map.

No amalgamation change: tests only.

• The changes are described in detail, both the what and why.
• If applicable, an existing issue is referenced.
• The Code coverage remained at 100%. A test case for every new line of code.
• If applicable, the documentation is updated.
• The source code is amalgamated by running make amalgamate. (n/a — test-only)

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[nlohmann]

Issue #5421 also mentions testing std::format. Please add this test or remove the "fixes #5421" from the description.

[nlohmann]

Adding to the earlier review — std::format is not the only gap against #5421. Beyond it:

Binary formats. The issue asks for a matrix over {ordered_json, alt_json} × {cbor, msgpack, ubjson, bson}. This PR covers ordered_json × four formats but BJData is missing, and alt_json is absent entirely — which is the half most likely to break, since the binary writers call .size() and iterate string_t byte-wise.

Also still untested with ordered_json:

• duplicate-key handling in a binary map (the issue calls this out specifically, since ordered_map::emplace differs from std::map's)
• merge_patch()
• at / value / contains with json_pointer

Other specializations from the issue, both untouched: the custom-allocator basic_json failure harness never reaching the binary readers, and alt_json with update(), merge_patch(), unflatten(), emplace/emplace_back, std::hash, three-way comparison.

The coverage that is here checks out — I ran all ten new assertions against develop and they pass, including the flatten() key order and the update(merge_objects=true) ordering.

---

Generated by Claude Code

[nlohmann]

@22elix3r My review comment from Aug 27 hasn't been addressed yet: issue #5421 also asks for a test of std::format — could you add it, or drop the "fixes #5421" from the description?

[nlohmann]

Thanks for this — good coverage of binary-format round trips and flatten/diff/patch for ordered_json.

We landed a more extensive version in #5480: same behaviors plus BJData (in addition to CBOR/MessagePack/UBJSON/BSON), an alt_json (custom string_t) sweep across all five binary formats, explicit key-order-preservation checks via iteration order (not just dump() string equality), a duplicate-key binary-decode regression test, merge_patch() coverage, and a std::format instantiation check. We're moving forward with #5480 and closing this one. Appreciate the contribution.

— closed by Claude Code on behalf of @nlohmann