[hhan2024]

Summary

Add shared parametric system tests for the contract that SDKs expose the selected UFC split's serialId in flag-evaluation metadata when span enrichment is enabled. The tests inspect the SDK's evaluation details through the parametric adapter; adapters must forward actual SDK metadata rather than reconstruct serial IDs from the configuration.

This PR introduces the tests and intentionally marks all eight language manifests (cpp, dotnet, golang, java, nodejs, php, python, ruby) as missing_feature. The SDK-specific PRs below expose metadata through the adapters, identify any SDK implementation gaps, and enable the tests at verified versions.

Contract

1. Flag-evaluation details always expose flagMetadata. When there is no metadata, the field may be an empty map or null; the field itself must not be omitted.
2. When span enrichment is enabled and the selected UFC split has a serial ID, that ID must appear in flag metadata. Positive IDs and zero are valid; zero must not be dropped by a truthiness check. A null or empty metadata map does not satisfy this case.
3. When the selected split has no serial ID, the serial-ID key must be absent. A key with a null value does not count as omission. With span enrichment enabled, a missing UFC serial ID is the only case where the serial-ID key is expected to be absent. Other metadata may still be present.
4. The ID must match the split actually selected by targeting and remain consistent across repeated evaluations of the same configuration and context.
5. .NET is the only representation exception: it returns the serial ID as an exact decimal string (for example, "42" or "0"); all other SDKs return a numeric value. .NET uses strings to preserve Int64 serial IDs through an OpenFeature metadata API whose integer getter only supports Int32. No response coercion is used to make the test pass.

The contract does not require serial-ID metadata when span enrichment is disabled.

The tests use each SDK's metadata key: Go uses dd.serialId; the other server SDKs use __dd_split_serial_id. The Go key handling and .NET string exception are implemented in their respective stacked PRs below, not in this parent branch alone.

Implementation

• Add tests/parametric/test_ffe/test_serial_id_metadata.py with five cases: positive ID, zero ID, missing-ID key omission, targeting-selected IDs, and repeated evaluations.
• Add serial-id-metadata-flags.json with fixtures for those cases, including separate targeted allocations with distinct serial IDs.
• Explicitly enable flagging and span enrichment via DD_EXPERIMENTAL_FLAGGING_PROVIDER_ENABLED=true and DD_EXPERIMENTAL_FLAGGING_PROVIDER_SPAN_ENRICHMENT_ENABLED=true.
• Deliver the UFC fixture through Remote Config, wait for acknowledgement, and retry evaluations while the provider is not ready.
• Add the feature_flags_serial_id_metadata feature marker (NOT_REPORTED_ID pending an assigned feature-parity ID).
• Use the approved utils.pytest facade, including the upstream facade on this branch so both PR-head and merged-tree imports work.

Server SDK test-enablement PRs

All seven system-tests PRs are stacked directly on this PR's branch, ffe-serial-id-metadata-system-test, rather than on one another. Version floors below are locally verified floors, not claims that every earlier release fails.

Server SDK	System-tests PR	Enablement / dependency
Go	#7844	Enables from v2.10.1; forwards SDK metadata and checks the native dd.serialId key. No SDK change required.
Node.js	#7852	Enables from v6.18.0; uses evaluation-detail APIs and forwards metadata. No SDK change required.
Python	#7853	Enables from v4.15.2; forwards SDK metadata. No SDK change required.
Java	#7854	Enables from v1.66.0; uses evaluation-detail APIs and forwards metadata. No SDK change required.
.NET	#7858	Enables from v3.54.0; adds FFE endpoints and the .NET-only string expectation. No SDK change required.
Ruby	#7856	Draft; forwards metadata and depends on dd-trace-rb #6400. Retains missing_feature until the SDK fix is merged and its release version is known.
PHP	#7857	Draft; preserves metadata as a JSON map and depends on dd-trace-php #4247. Retains missing_feature until the SDK fix is merged and its release version is known.

C++ remains missing_feature; there is no C++ activation PR in this set.

Validation

• Parent test-module import and repository formatting/lint checks passed, including the merged-main import policy.
• All five serial-ID tests passed without force overrides in the Go, Node.js, Python, Java, and .NET activation branches at the versions above.
• Ruby and PHP candidate SDK fixes passed all five serial-ID cases locally; those tests are not yet enabled for released SDKs. Their SDK unit-test and lint evidence is recorded in the linked PRs.
• Node.js, Python, and Java also passed all 24 dynamic-evaluation regression cases. The .NET draft documents two separate JSON-default/no-allocation failures from that regression suite; these are not serial-ID failures and are not hidden by this contract exception.

Follow-up

• Assign a real feature-parity ID to replace NOT_REPORTED_ID.
• Merge the SDK-specific adapter/activation PRs alongside this shared test foundation.
• For Ruby and PHP, merge the SDK fixes and select their release floors before replacing missing_feature.

[github-actions]

CODEOWNERS have been resolved as:

tests/parametric/test_ffe/serial-id-metadata-flags.json                 @DataDog/feature-flagging-and-experimentation-sdk
tests/parametric/test_ffe/test_serial_id_metadata.py                    @DataDog/feature-flagging-and-experimentation-sdk
manifests/cpp.yml                                                       @DataDog/system-tests-reviewers
manifests/dotnet.yml                                                    @DataDog/system-tests-reviewers
manifests/golang.yml                                                    @DataDog/system-tests-reviewers
manifests/java.yml                                                      @DataDog/system-tests-reviewers
manifests/nodejs.yml                                                    @DataDog/system-tests-reviewers
manifests/php.yml                                                       @DataDog/system-tests-reviewers
manifests/python.yml                                                    @DataDog/system-tests-reviewers
manifests/ruby.yml                                                      @DataDog/system-tests-reviewers
manifests/rust.yml                                                      @DataDog/system-tests-reviewers
tests/parametric/test_ffe/test_dynamic_evaluation.py                    @DataDog/feature-flagging-and-experimentation-sdk
utils/_features.py                                                      @DataDog/system-tests-core

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 194539dc51

ℹ️ About Codex in GitHub

Codex has been enabled to automatically 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".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 9547cefa87

ℹ️ About Codex in GitHub

Codex has been enabled to automatically 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".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[leoromanovsky]

Nice!