feat(tracing)!: JSON-serialize propagated metadata values

[hassiebp]

What does this PR do?

propagate_attributes(metadata=...) coerced non-string values with str(), so True became "True", ["a"] became "['a']", and dicts became their Python repr. Propagated metadata values are now serialized exactly like JSON.stringify in the JS SDK (langfuse-js#980), so both SDKs emit byte-identical values: compact separators, non-ASCII kept as is, insertion key order, and None → "null".

value	before	after (= JS)
True	"True"	"true"
["branch:agent"]	"['branch:agent']"	'["branch:agent"]'
{"model": "gpt-4o", "n": {"a": "ü"}}	"{'model': 'gpt-4o', 'n': {'a': 'ü'}}"	'{"model":"gpt-4o","n":{"a":"ü"}}'
None	"None"	"null"

Strings are kept as is, and ints serialize the same as before. The 200-character limit applies to the serialized value. Non-JSON Python types (datetimes, pydantic models, and so on) go through EventSerializer, as elsewhere in the SDK.

Stack: 5 of 10. Base: #1931. Merge after #1931.

Type of change

• Bug fix
• New feature
• Breaking change
• Refactor
• Documentation update
• Tooling, CI, or repo maintenance

Verification

uv run --frozen ruff check .                       # All checks passed!
uv run --frozen mypy langfuse --no-error-summary   # clean
uv run --frozen pytest -q tests/unit/test_propagate_attributes.py  # the new test failed before the change; 133 passed after
uv run --frozen pytest -n auto tests/unit          # 714 passed, 2 skipped

Byte-equality with Node's JSON.stringify was checked for ints, floats, bools, null, lists, nested dicts, non-ASCII/emoji, escaped quotes and tabs, and large safe ints: all identical.

Notes for the reviewer

These Python-only cases can't match JS exactly, by design:

• Integral floats: 1.0 serializes as "1.0" here, but "1" in JS.
• NaN/Infinity: these become "\"NaN\"" here, but "null" in JS.
• Datetimes: they use EventSerializer's ISO format (no forced milliseconds), so they aren't identical to JS Date.toJSON.
• Tuples and sets: these serialize as JSON arrays.

Checklist

• I self-reviewed the diff using code_review.md.
• I added or updated tests for behavior changes.
• I updated docs, examples, or .env.template if needed.
• I did not hand-edit generated files; if generated files changed, I used the upstream regeneration path.
• I did not commit secrets or credentials.

  

Confidence Score: 4/5

The PR should not merge until propagated large integer metadata retains its intended representation.

Summary

The PR JSON-serializes non-string propagated metadata, drops None values, and updates unit tests.

• Large integers acquire JSON quotation marks in propagated attributes.
• The tests do not cover a baggage-header round trip for the new value formats.

_{Reviews (1) · Last reviewed commit: "feat(tracing)!: JSON-serialize propagate..."}

[github-actions]

@claude review

[claude]

Claude Code Review

No review was started: this request came from a bot account. Manual reviews can only be requested by someone with write access to this repository. Ask a maintainer to comment @claude review, or have your automation post the comment from a user account with write access.

_{Tip: disable this comment in your organization's Code Review settings.}

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-06T14:34:16.155740Z	e9623ee	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[claude]

Nothing blocking. The comments below are optional suggestions. There is no need to push a fix for them before merging.