Do not emit BJData Draft-3 byte marker in Draft 2

[22elix3r]

Summary

`to_bjdata()` mapped ndarray `ArrayType = "byte"` to marker `B` even with the default Draft-2 version. `B` is Draft-3-only, so that stream is invalid for Draft-2 readers and does not round-trip as an annotated array.

Related Issue

Fixes #5404

Changes Made

• If the ndarray type is `byte` and the version is not Draft 3, fall back to plain object encoding (`write_bjdata_ndarray` returns true)
• Test that Draft 2 output contains no `B` and parses as an object; Draft 3 still emits `B`

Testing

Commands executed:

• `python3 tools/amalgamate/amalgamate.py -c tools/amalgamate/config_json.json -s .`
• `cmake -S . -B build -DJSON_BuildTests=ON -DCMAKE_BUILD_TYPE=Debug`
• `cmake --build build --target test-bjdata_cpp11`
• `./tests/test-bjdata_cpp11 --no-skip -tce='downloaded' -tc='5404'`

Results:

• 1 passed, 3 assertions

Notes

uint8 remains the Draft-2 way to encode 8-bit values. Draft 3 is unchanged.

• 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`.

[Copilot]

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

[nlohmann]

The Draft-2 gate works — I checked the bytes on this branch:

draft2: 7b ... 5f 41 72 72 61 79 54 79 70 65 5f ...   -> plain object, no 'B'
draft3: 5b 24 42 23 5b 69 02 5d 01 02                 -> '[$B#[i2]..' as before

Two things from #5404 are still open, though.

1. The second half of the issue. Its title is "and the value does not round-trip", and the body raises it as a separate defect:

Separately, unlike every other _ArrayType_ (e.g. uint8, which round-trips back to the annotated object), the result parses back as a binary value, so to_bjdata/from_bjdata changes the value.

That is unchanged in Draft 3 — json::from_bjdata(json::to_bjdata(j, false, false, draft3)) still gives {"bytes":[1,2],"subtype":null} rather than the original annotation object. The PR's own Draft-3 assertion only greps for a B byte and doesn't check the round trip. Fine to declare that inherent and out of scope, but then it should be said so on the issue rather than closed by this PR.

2. Docs. docs/mkdocs/docs/features/binary_formats/bjdata.md:127 still lists the accepted _ArrayType_ values as

double, char, or byte,

with no Draft-3 caveat, even though byte now silently falls back to object encoding under the default version. Compare #5438, which does update this same paragraph for its change.

---

Generated by Claude Code

[github-actions]

🔴 Amalgamation check failed! 🔴

The source code has not been amalgamated and/or formatted correctly.

📎 A ready-to-apply patch is attached to the failed workflow run as the amalgamation-patch artifact. Download it, then apply it locally from the repository root with:

git apply amalgamation.patch

This does not require installing astyle yourself.