Support scoped attributes in test quarantine workflow

[wtgodbe]

Support scoped attributes in test quarantine workflow

• You've read the Contributor Guide and Code of Conduct.
• You've included unit or integration tests for your change, where applicable.
• You've included inline docs for your change, where applicable.
• There's an open issue for the PR that you are making. If you'd like to propose a new feature or change, please open an issue to discuss the change or find an existing issue.

Support scoped quarantine attributes

Description

Updates the test-quarantine workflow for the attribute changes introduced by #69584:

• Preserves exact theory-row identities in failure evidence and correlates unambiguous InlineData rows with QuarantinedTestData.
• Tracks row-level quarantine history and validates exact InlineData/QuarantinedTestData transformations at the safe-output boundary.
• Supports operating-system-scoped QuarantinedTest and QuarantinedTestData additions through deterministic, receipt-bound OS scopes, falling back to all supported platforms when the evidence is incomplete.
• Keeps multiline and combined attribute sections visible to quarantine-state/history collection while limiting automated row rewrites to one-line attributes.
• Narrows POST_ClientCancellationBidirectional_RequestAbortRaised to quarantine only its demonstrated failing HTTP/3 row; the HTTP/2 row remains enabled.

Validation:

• node .github/workflows/scripts/test-quarantine/test_kbe_issue_handler.js — passed.
• python3 -B .github/workflows/scripts/test-quarantine/test_collect_case_a_eligibility.py — passed.
• python3 -B .github/workflows/scripts/test-quarantine/test_validate_pull_request_outputs.py — passed.
• gh aw compile test-quarantine --no-emit --strict — passed with no warnings.
• gh aw compile test-quarantine — passed and regenerated .github/workflows/test-quarantine.lock.yml.
• source activate.sh && ./src/Servers/build.sh --projects "$(pwd)/src/Servers/Kestrel/test/Interop.FunctionalTests/Interop.FunctionalTests.csproj" --no-build-native --no-build-nodejs --no-build-java --no-build-installers — passed with 0 warnings and 0 errors.

Related to #69584

[wtgodbe]

CC @javiercn

[Copilot]

Copilot review overview

🟡 Changes recommended

Row rewrites can break ordinary xUnit theories, and multi-build failures incorrectly fall back to all-platform quarantine.

Review effort: Balanced
Findings: 1 · 1

Open (2)

• Restrict matching inline data to ConditionalTheory methods · New
• Preserve per-build OS identity for all retained failures · New

What changed in this PR

Extends test quarantine automation with row-level and OS-scoped quarantines.

Changes:

• Preserves theory-row identities and adds QuarantinedTestData handling.
• Adds deterministic OS scopes and safe-output validation.
• Narrows one Kestrel quarantine to its HTTP/3 row.

File	Description
Http3RequestTests.cs	Quarantines only the HTTP/3 row.
test-quarantine.md	Updates workflow logic and instructions.
test-quarantine.lock.yml	Regenerates the compiled workflow.
validate_pull_request_outputs.py	Validates row and OS transformations.
test_validate_pull_request_outputs.py	Tests safe-output validation.
test_collect_case_a_eligibility.py	Tests row eligibility and OS scopes.
README.md	Documents new safety properties.
collect_case_a_eligibility.py	Adds row resolution, history, and OS collection.

---

💡 Add a code-review agent skill for context-aware, tailored reviews. Learn more in the docs.

[Copilot]

Copilot review overview

🟡 Changes recommended

Platform inference uses OS-neutral Helix work-item names, causing scoped quarantines to fall back to all platforms.

Review effort: Balanced
Findings: 1

Open (1)

• Derive quarantine OS from Helix QueueName, not work item name · New

Resolved since last review (2)

• Restrict matching inline data to ConditionalTheory methods
• Preserve per-build OS identity for all retained failures

[Copilot]

Copilot review overview

🟡 Changes recommended

Row matching mishandles common literals and trailing comments, while renamed row history can incorrectly permit automated unquarantine.

Review effort: Balanced
Findings: 1

Open (1)

• Normalize inline data values to xUnit-rendered arguments · New

Resolved since last review (1)

• Derive quarantine OS from Helix QueueName, not work item name

[Copilot]

Copilot review overview

🟡 Changes recommended

Commented theory rows are marked eligible but their generated patches are always rejected by safe-output validation.

Review effort: Balanced
Findings: 1

Open (1)

• Safe-output gate rejects eligible commented theory rows · New

Resolved since last review (1)

• Normalize inline data values to xUnit-rendered arguments

[Copilot]

Copilot review overview

🟡 Changes recommended

Mixed inline and external theory data can cause the workflow to quarantine the wrong row.

Review effort: Balanced
Findings: 1

Open (1)

• Avoid misattributing failures to matching InlineData rows · New

Resolved since last review (1)

• Safe-output gate rejects eligible commented theory rows

[Copilot]

Copilot review overview

🟡 Changes recommended

The collector treats MsQuicSupported as a potential data provider, preventing the newly scoped Kestrel row from being resolved or later unquarantined.

Review effort: Balanced
Findings: 1

Open (1)

• Avoid treating non-data attributes as ambiguous data providers · New

Resolved since last review (1)

• Avoid misattributing failures to matching InlineData rows

[Copilot]

Copilot review overview

🔵 Needs a closer look

It changes privileged workflow evidence collection and patch validation across external Azure DevOps and Helix data paths that source inspection cannot fully verify.

Review effort: Balanced
Findings: None

Resolved since last review (1)

• Avoid treating non-data attributes as ambiguous data providers