[perf-improver] perf: avoid List(TestResult) allocation in RunTestMethodAsync non-data-driven fast path

[Amaury Levé (Evangelink)]

Goal and Rationale

For the vast majority of test executions — non-data-driven tests — RunTestMethodAsync was paying three unnecessary costs on every call:

1. Two attribute scans: TryExecuteDataSourceBasedTestsAsync scans all cached method attributes for DataSourceAttribute; if that returns false, TryExecuteFoldedDataDrivenTestsAsync scans them again for ITestDataSource implementors.
2. List<TestResult> allocation: a List<TestResult> is created even though only ExecuteTestAsync will be called (which already returns TestResult[]).
3. Spread allocation: return [.. results] converts the list back to an array, allocating a second short-lived TestResult[].

In a typical suite with 1,000 non-data-driven tests, this adds ~80 KB of short-lived allocations and ~2,000 extra attribute-scan iterations per run.

Approach

New IsDataDrivenTest() private method: one pass over GetCustomAttributesCached that checks attribute is DataSourceAttribute or UTF.ITestDataSource — combining both scans into one.

Fast path (new): if _test.DataType != DynamicDataType.ITestDataSourceandIsDataDrivenTest() returns false, the method:

• Calls ExecuteTestAsync directly
• Returns its TestResult[] without allocating a List<TestResult>
• Performs one attribute-cache scan instead of two

Slow path (unchanged): all data-driven tests continue through the existing List<TestResult> path.

GetAggregateOutcome signature: widened from List<TestResult> to IReadOnlyList<TestResult> so both the fast-path TestResult[] and slow-path List<TestResult> can call it (arrays and lists both implement IReadOnlyList<T>).

Performance Evidence

Per non-data-driven test, the change eliminates approximately:

• 1 List<TestResult> object (~56 bytes: 24-byte header + internal capacity-4 backing array)
• 1 TestResult[] spread allocation (~24 bytes, length 1)
• 1 redundant attribute-cache enumeration (second TryExecute... scan avoided)

For a 1,000-test non-data-driven suite: ~80 KB fewer short-lived allocations and ~2,000 fewer attribute-scan iterations per run.

No local SDK is available in this CI agent; build/test verification is delegated to CI.

Trade-offs

• Very slight code duplication: the "set display name → execute → set outcome" block appears in both the fast path and the slow path's else branch. The else branch is an edge case (test has DataSourceAttribute but multiple such attributes cause TryExecuteDataSourceBasedTestsAsync to skip it), so it is never exercised on common paths.
• No behavioral changes: the fast path is only reached when both DataType != ITestDataSource and IsDataDrivenTest() is false, ensuring data-driven tests are unaffected.

Test Status

Delegated to CI. Existing unit tests in TestMethodRunnerTests.cs exercise both the new fast path (non-data-driven) and the unchanged slow path (data-driven) code paths.

Reproducibility

# Build and run adapter unit tests
./build.sh -test

🤖 Automated content by GitHub Copilot. Posted via a maintainer's GitHub token, so it appears under their account — the account owner did not write or approve this content personally. Generated by the Perf Improver workflow. · 1.7K AIC · ⌖ 26.2 AIC · ⊞ 57.7K · [◷]( · ◷)

Add this agentic workflows to your repo

To install this agentic workflow, run

gh aw add githubnext/agentics/workflows/perf-improver.md@main

[Copilot]

Pull request overview

This PR optimizes MSTest adapter execution for the common non-data-driven case by avoiding redundant attribute scans and short-lived allocations in TestMethodRunner.RunTestMethodAsync.

Changes:

• Added a private IsDataDrivenTest() helper to detect DataSourceAttribute / ITestDataSource in a single cached-attribute pass.
• Introduced a non-data-driven fast path that calls ExecuteTestAsync directly and returns its TestResult[] without allocating/spreading a List<TestResult>.
• Widened GetAggregateOutcome from List<TestResult> to IReadOnlyList<TestResult> to support both arrays and lists.

Show a summary per file

File	Description
src/Adapter/MSTestAdapter.PlatformServices/Execution/TestMethodRunner.cs	Adds a non-data-driven fast path and consolidates attribute scanning to reduce allocations and iterations.

Review details

• Files reviewed: 1/1 changed files
• Comments generated: 1
• Review effort level: Low

[Amaury Levé (Evangelink)]

Note

🤖 Automated review by GitHub Copilot. Posted via a maintainer's GitHub token, so it appears under their account — the account owner did not write or approve this content personally. Generated by the Expert Code Review workflow. To request a follow-up action, reply by tagging @copilot directly.

Review Summary

Reviewed src/Adapter/MSTestAdapter.PlatformServices/Execution/TestMethodRunner.cs against all 22 applicable dimensions. The optimization is well-reasoned and algorithmically sound for the common case, but there is one correctness concern worth addressing before merge.

---

Verdict Table

#	Dimension	Status	Notes
1	Algorithmic Correctness	⚠️SEE INLINE	Fast path bypasses results.Count == 0 guard (detail below)
2	Threading & Concurrency	✅ LGTM	No new shared-state access; all reads are already synchronized by callers
3	Security & IPC Contract Safety	✅ LGTM	No security-relevant code touched
4	Public API & Binary Compatibility	✅ LGTM	All changes are private; IReadOnlyList<T> widening is private-only
5	Performance & Allocations	✅ LGTM	Correctly eliminates List<TestResult> + spread allocation on the hot path
6	Cross-TFM Compatibility	✅ LGTM	No TFM-conditional APIs introduced
7	Resource & IDisposable Management	✅ LGTM	No disposables affected
8	Defensive Coding at Boundaries	⚠️ Same as #1	See inline comment on line 134
9	Localization & Resources	✅ LGTM	No new user-facing strings added
10	Test Isolation	✅ LGTM	No static mutation introduced
11	Assertion Quality	✅ LGTM	Tests use AwesomeAssertions (project convention)
12	Flakiness Patterns	✅ LGTM	No timing/ordering sensitivity introduced
13	Test Completeness & Coverage	✅ LGTM	Existing non-data-driven tests exercise the fast path; data-driven tests cover the slow path
14	Data-Driven Test Coverage	N/A	No new data-driven tests
15	Code Structure & Simplification	✅ LGTM	Duplication acknowledged in PR; else branch is an edge case only
16	Naming & Conventions	✅ LGTM	IsDataDrivenTest() name is clear and consistent with codebase style
17	Documentation Accuracy	✅ LGTM	XML doc and inline comments are accurate
18	Analyzer & Code Fix Quality	N/A	No src/Analyzers/ changes
19	IPC Wire Compatibility	N/A	No serialization code changed
20	Build Infrastructure & Dependencies	✅ LGTM	No build changes
21	Scope & PR Discipline	✅ LGTM	Focused, single-file, single-concern change
22	PowerShell Scripting Hygiene	N/A	No .ps1 files changed

---

The One Correctness Issue (inline comment, line 134)

DynamicDataType has only two values (None, ITestDataSource), and IsDataDrivenTest() correctly mirrors the exact attribute checks performed by TryExecuteDataSourceBasedTestsAsync (looks for DataSourceAttribute) and TryExecuteFoldedDataDrivenTestsAsync (looks for UTF.ITestDataSource). The fast-path gate is logically equivalent to the old else branch and the algorithm is correct.

The single gap: the slow path has a defensive guard at lines 190-199 that ensures RunTestMethodAsync never returns [] — because ExecuteAsync's finally block unconditionally dereferences result![0], an empty return would throw IndexOutOfRangeException rather than surfacing a clean error. The fast path omits this guard.

For all practical purposes (standard TestMethodAttribute always returns ≥ 1 result) this is benign. But a custom executor returning TestResult[0]{} would now crash on the fast path where it previously received a graceful error result. The inline comment explains the fix options.

No other issues found across the remaining 20 applicable dimensions.

[Amaury Levé (Evangelink)]

Goal and Rationale

For the vast majority of test executions — non-data-driven tests — RunTestMethodAsync was paying three unnecessary costs on every call:

1. Two attribute scans: TryExecuteDataSourceBasedTestsAsync scans all cached method attributes for DataSourceAttribute; if that returns false, TryExecuteFoldedDataDrivenTestsAsync scans them again for ITestDataSource implementors.
2. List<TestResult> allocation: a List<TestResult> is created even though only ExecuteTestAsync will be called (which already returns TestResult[]).
3. Spread allocation: return [.. results] converts the list back to an array, allocating a second short-lived TestResult[].

In a typical suite with 1,000 non-data-driven tests, this adds ~80 KB of short-lived allocations and ~2,000 extra attribute-scan iterations per run.

Approach

New IsDataDrivenTest() private method: one pass over GetCustomAttributesCached that checks attribute is DataSourceAttribute or UTF.ITestDataSource — combining both scans into one.

Fast path (new): if _test.DataType != DynamicDataType.ITestDataSourceandIsDataDrivenTest() returns false, the method:

• Calls ExecuteTestAsync directly
• Returns its TestResult[] without allocating a List<TestResult>
• Performs one attribute-cache scan instead of two

Slow path (unchanged): all data-driven tests continue through the existing List<TestResult> path.

GetAggregateOutcome signature: widened from List<TestResult> to IReadOnlyList<TestResult> so both the fast-path TestResult[] and slow-path List<TestResult> can call it (arrays and lists both implement IReadOnlyList<T>).

Performance Evidence

Per non-data-driven test, the change eliminates approximately:

• 1 List<TestResult> object (~56 bytes: 24-byte header + internal capacity-4 backing array)
• 1 TestResult[] spread allocation (~24 bytes, length 1)
• 1 redundant attribute-cache enumeration (second TryExecute... scan avoided)

For a 1,000-test non-data-driven suite: ~80 KB fewer short-lived allocations and ~2,000 fewer attribute-scan iterations per run.

No local SDK is available in this CI agent; build/test verification is delegated to CI.

Trade-offs

• Very slight code duplication: the "set display name → execute → set outcome" block appears in both the fast path and the slow path's else branch. The else branch is an edge case (test has DataSourceAttribute but multiple such attributes cause TryExecuteDataSourceBasedTestsAsync to skip it), so it is never exercised on common paths.
• No behavioral changes: the fast path is only reached when both DataType != ITestDataSource and IsDataDrivenTest() is false, ensuring data-driven tests are unaffected.

Test Status

Delegated to CI. Existing unit tests in TestMethodRunnerTests.cs exercise both the new fast path (non-data-driven) and the unchanged slow path (data-driven) code paths.

Reproducibility

# Build and run adapter unit tests
./build.sh -test

🤖 Automated content by GitHub Copilot. Posted via a maintainer's GitHub token, so it appears under their account — the account owner did not write or approve this content personally. Generated by the Perf Improver workflow. · 1.7K AIC · ⌖ 26.2 AIC · ⊞ 57.7K · [◷]( · ◷)

Add this agentic workflows to your repo

To install this agentic workflow, run

gh aw add githubnext/agentics/workflows/perf-improver.md@main

[Copilot]

Pull request overview

This PR optimizes MSTest adapter execution for the common non-data-driven case by avoiding redundant attribute scans and short-lived allocations in TestMethodRunner.RunTestMethodAsync.

Changes:

• Added a private IsDataDrivenTest() helper to detect DataSourceAttribute / ITestDataSource in a single cached-attribute pass.
• Introduced a non-data-driven fast path that calls ExecuteTestAsync directly and returns its TestResult[] without allocating/spreading a List<TestResult>.
• Widened GetAggregateOutcome from List<TestResult> to IReadOnlyList<TestResult> to support both arrays and lists.

Show a summary per file

File	Description
src/Adapter/MSTestAdapter.PlatformServices/Execution/TestMethodRunner.cs	Adds a non-data-driven fast path and consolidates attribute scanning to reduce allocations and iterations.

Review details

• Files reviewed: 1/1 changed files
• Comments generated: 1
• Review effort level: Low

[Amaury Levé (Evangelink)]

Note

🤖 Automated review by GitHub Copilot. Posted via a maintainer's GitHub token, so it appears under their account — the account owner did not write or approve this content personally. Generated by the Expert Code Review workflow. To request a follow-up action, reply by tagging @copilot directly.

Review Summary

Reviewed src/Adapter/MSTestAdapter.PlatformServices/Execution/TestMethodRunner.cs against all 22 applicable dimensions. The optimization is well-reasoned and algorithmically sound for the common case, but there is one correctness concern worth addressing before merge.

---

Verdict Table

#	Dimension	Status	Notes
1	Algorithmic Correctness	⚠️SEE INLINE	Fast path bypasses results.Count == 0 guard (detail below)
2	Threading & Concurrency	✅ LGTM	No new shared-state access; all reads are already synchronized by callers
3	Security & IPC Contract Safety	✅ LGTM	No security-relevant code touched
4	Public API & Binary Compatibility	✅ LGTM	All changes are private; IReadOnlyList<T> widening is private-only
5	Performance & Allocations	✅ LGTM	Correctly eliminates List<TestResult> + spread allocation on the hot path
6	Cross-TFM Compatibility	✅ LGTM	No TFM-conditional APIs introduced
7	Resource & IDisposable Management	✅ LGTM	No disposables affected
8	Defensive Coding at Boundaries	⚠️ Same as #1	See inline comment on line 134
9	Localization & Resources	✅ LGTM	No new user-facing strings added
10	Test Isolation	✅ LGTM	No static mutation introduced
11	Assertion Quality	✅ LGTM	Tests use AwesomeAssertions (project convention)
12	Flakiness Patterns	✅ LGTM	No timing/ordering sensitivity introduced
13	Test Completeness & Coverage	✅ LGTM	Existing non-data-driven tests exercise the fast path; data-driven tests cover the slow path
14	Data-Driven Test Coverage	N/A	No new data-driven tests
15	Code Structure & Simplification	✅ LGTM	Duplication acknowledged in PR; else branch is an edge case only
16	Naming & Conventions	✅ LGTM	IsDataDrivenTest() name is clear and consistent with codebase style
17	Documentation Accuracy	✅ LGTM	XML doc and inline comments are accurate
18	Analyzer & Code Fix Quality	N/A	No src/Analyzers/ changes
19	IPC Wire Compatibility	N/A	No serialization code changed
20	Build Infrastructure & Dependencies	✅ LGTM	No build changes
21	Scope & PR Discipline	✅ LGTM	Focused, single-file, single-concern change
22	PowerShell Scripting Hygiene	N/A	No .ps1 files changed

---

The One Correctness Issue (inline comment, line 134)

DynamicDataType has only two values (None, ITestDataSource), and IsDataDrivenTest() correctly mirrors the exact attribute checks performed by TryExecuteDataSourceBasedTestsAsync (looks for DataSourceAttribute) and TryExecuteFoldedDataDrivenTestsAsync (looks for UTF.ITestDataSource). The fast-path gate is logically equivalent to the old else branch and the algorithm is correct.

The single gap: the slow path has a defensive guard at lines 190-199 that ensures RunTestMethodAsync never returns [] — because ExecuteAsync's finally block unconditionally dereferences result![0], an empty return would throw IndexOutOfRangeException rather than surfacing a clean error. The fast path omits this guard.

For all practical purposes (standard TestMethodAttribute always returns ≥ 1 result) this is benign. But a custom executor returning TestResult[0]{} would now crash on the fast path where it previously received a graceful error result. The inline comment explains the fix options.

No other issues found across the remaining 20 applicable dimensions.

[Amaury Levé (Evangelink)]

Goal and Rationale

For the vast majority of test executions — non-data-driven tests — RunTestMethodAsync was paying three unnecessary costs on every call:

1. Two attribute scans: TryExecuteDataSourceBasedTestsAsync scans all cached method attributes for DataSourceAttribute; if that returns false, TryExecuteFoldedDataDrivenTestsAsync scans them again for ITestDataSource implementors.
2. List<TestResult> allocation: a List<TestResult> is created even though only ExecuteTestAsync will be called (which already returns TestResult[]).
3. Spread allocation: return [.. results] converts the list back to an array, allocating a second short-lived TestResult[].

In a typical suite with 1,000 non-data-driven tests, this adds ~80 KB of short-lived allocations and ~2,000 extra attribute-scan iterations per run.

Approach

New IsDataDrivenTest() private method: one pass over GetCustomAttributesCached that checks attribute is DataSourceAttribute or UTF.ITestDataSource — combining both scans into one.

Fast path (new): if _test.DataType != DynamicDataType.ITestDataSourceandIsDataDrivenTest() returns false, the method:

• Calls ExecuteTestAsync directly
• Returns its TestResult[] without allocating a List<TestResult>
• Performs one attribute-cache scan instead of two

Slow path (unchanged): all data-driven tests continue through the existing List<TestResult> path.

GetAggregateOutcome signature: widened from List<TestResult> to IReadOnlyList<TestResult> so both the fast-path TestResult[] and slow-path List<TestResult> can call it (arrays and lists both implement IReadOnlyList<T>).

Performance Evidence

Per non-data-driven test, the change eliminates approximately:

• 1 List<TestResult> object (~56 bytes: 24-byte header + internal capacity-4 backing array)
• 1 TestResult[] spread allocation (~24 bytes, length 1)
• 1 redundant attribute-cache enumeration (second TryExecute... scan avoided)

For a 1,000-test non-data-driven suite: ~80 KB fewer short-lived allocations and ~2,000 fewer attribute-scan iterations per run.

No local SDK is available in this CI agent; build/test verification is delegated to CI.

Trade-offs

• Very slight code duplication: the "set display name → execute → set outcome" block appears in both the fast path and the slow path's else branch. The else branch is an edge case (test has DataSourceAttribute but multiple such attributes cause TryExecuteDataSourceBasedTestsAsync to skip it), so it is never exercised on common paths.
• No behavioral changes: the fast path is only reached when both DataType != ITestDataSource and IsDataDrivenTest() is false, ensuring data-driven tests are unaffected.

Test Status

Delegated to CI. Existing unit tests in TestMethodRunnerTests.cs exercise both the new fast path (non-data-driven) and the unchanged slow path (data-driven) code paths.

Reproducibility

# Build and run adapter unit tests
./build.sh -test

🤖 Automated content by GitHub Copilot. Posted via a maintainer's GitHub token, so it appears under their account — the account owner did not write or approve this content personally. Generated by the Perf Improver workflow. · 1.7K AIC · ⌖ 26.2 AIC · ⊞ 57.7K · [◷]( · ◷)

Add this agentic workflows to your repo

To install this agentic workflow, run

gh aw add githubnext/agentics/workflows/perf-improver.md@main

[Copilot]

Pull request overview

This PR optimizes MSTest adapter execution for the common non-data-driven case by avoiding redundant attribute scans and short-lived allocations in TestMethodRunner.RunTestMethodAsync.

Changes:

• Added a private IsDataDrivenTest() helper to detect DataSourceAttribute / ITestDataSource in a single cached-attribute pass.
• Introduced a non-data-driven fast path that calls ExecuteTestAsync directly and returns its TestResult[] without allocating/spreading a List<TestResult>.
• Widened GetAggregateOutcome from List<TestResult> to IReadOnlyList<TestResult> to support both arrays and lists.

Show a summary per file

File	Description
src/Adapter/MSTestAdapter.PlatformServices/Execution/TestMethodRunner.cs	Adds a non-data-driven fast path and consolidates attribute scanning to reduce allocations and iterations.

Review details

• Files reviewed: 1/1 changed files
• Comments generated: 1
• Review effort level: Low

[Amaury Levé (Evangelink)]

Note

🤖 Automated review by GitHub Copilot. Posted via a maintainer's GitHub token, so it appears under their account — the account owner did not write or approve this content personally. Generated by the Expert Code Review workflow. To request a follow-up action, reply by tagging @copilot directly.

Review Summary

Reviewed src/Adapter/MSTestAdapter.PlatformServices/Execution/TestMethodRunner.cs against all 22 applicable dimensions. The optimization is well-reasoned and algorithmically sound for the common case, but there is one correctness concern worth addressing before merge.

---

Verdict Table

#	Dimension	Status	Notes
1	Algorithmic Correctness	⚠️SEE INLINE	Fast path bypasses results.Count == 0 guard (detail below)
2	Threading & Concurrency	✅ LGTM	No new shared-state access; all reads are already synchronized by callers
3	Security & IPC Contract Safety	✅ LGTM	No security-relevant code touched
4	Public API & Binary Compatibility	✅ LGTM	All changes are private; IReadOnlyList<T> widening is private-only
5	Performance & Allocations	✅ LGTM	Correctly eliminates List<TestResult> + spread allocation on the hot path
6	Cross-TFM Compatibility	✅ LGTM	No TFM-conditional APIs introduced
7	Resource & IDisposable Management	✅ LGTM	No disposables affected
8	Defensive Coding at Boundaries	⚠️ Same as #1	See inline comment on line 134
9	Localization & Resources	✅ LGTM	No new user-facing strings added
10	Test Isolation	✅ LGTM	No static mutation introduced
11	Assertion Quality	✅ LGTM	Tests use AwesomeAssertions (project convention)
12	Flakiness Patterns	✅ LGTM	No timing/ordering sensitivity introduced
13	Test Completeness & Coverage	✅ LGTM	Existing non-data-driven tests exercise the fast path; data-driven tests cover the slow path
14	Data-Driven Test Coverage	N/A	No new data-driven tests
15	Code Structure & Simplification	✅ LGTM	Duplication acknowledged in PR; else branch is an edge case only
16	Naming & Conventions	✅ LGTM	IsDataDrivenTest() name is clear and consistent with codebase style
17	Documentation Accuracy	✅ LGTM	XML doc and inline comments are accurate
18	Analyzer & Code Fix Quality	N/A	No src/Analyzers/ changes
19	IPC Wire Compatibility	N/A	No serialization code changed
20	Build Infrastructure & Dependencies	✅ LGTM	No build changes
21	Scope & PR Discipline	✅ LGTM	Focused, single-file, single-concern change
22	PowerShell Scripting Hygiene	N/A	No .ps1 files changed

---

The One Correctness Issue (inline comment, line 134)

DynamicDataType has only two values (None, ITestDataSource), and IsDataDrivenTest() correctly mirrors the exact attribute checks performed by TryExecuteDataSourceBasedTestsAsync (looks for DataSourceAttribute) and TryExecuteFoldedDataDrivenTestsAsync (looks for UTF.ITestDataSource). The fast-path gate is logically equivalent to the old else branch and the algorithm is correct.

The single gap: the slow path has a defensive guard at lines 190-199 that ensures RunTestMethodAsync never returns [] — because ExecuteAsync's finally block unconditionally dereferences result![0], an empty return would throw IndexOutOfRangeException rather than surfacing a clean error. The fast path omits this guard.

For all practical purposes (standard TestMethodAttribute always returns ≥ 1 result) this is benign. But a custom executor returning TestResult[0]{} would now crash on the fast path where it previously received a graceful error result. The inline comment explains the fix options.

No other issues found across the remaining 20 applicable dimensions.

[Amaury Levé (Evangelink)]

Goal and Rationale

For the vast majority of test executions — non-data-driven tests — RunTestMethodAsync was paying three unnecessary costs on every call:

1. Two attribute scans: TryExecuteDataSourceBasedTestsAsync scans all cached method attributes for DataSourceAttribute; if that returns false, TryExecuteFoldedDataDrivenTestsAsync scans them again for ITestDataSource implementors.
2. List<TestResult> allocation: a List<TestResult> is created even though only ExecuteTestAsync will be called (which already returns TestResult[]).
3. Spread allocation: return [.. results] converts the list back to an array, allocating a second short-lived TestResult[].

In a typical suite with 1,000 non-data-driven tests, this adds ~80 KB of short-lived allocations and ~2,000 extra attribute-scan iterations per run.

Approach

New IsDataDrivenTest() private method: one pass over GetCustomAttributesCached that checks attribute is DataSourceAttribute or UTF.ITestDataSource — combining both scans into one.

Fast path (new): if _test.DataType != DynamicDataType.ITestDataSourceandIsDataDrivenTest() returns false, the method:

• Calls ExecuteTestAsync directly
• Returns its TestResult[] without allocating a List<TestResult>
• Performs one attribute-cache scan instead of two

Slow path (unchanged): all data-driven tests continue through the existing List<TestResult> path.

GetAggregateOutcome signature: widened from List<TestResult> to IReadOnlyList<TestResult> so both the fast-path TestResult[] and slow-path List<TestResult> can call it (arrays and lists both implement IReadOnlyList<T>).

Performance Evidence

Per non-data-driven test, the change eliminates approximately:

• 1 List<TestResult> object (~56 bytes: 24-byte header + internal capacity-4 backing array)
• 1 TestResult[] spread allocation (~24 bytes, length 1)
• 1 redundant attribute-cache enumeration (second TryExecute... scan avoided)

For a 1,000-test non-data-driven suite: ~80 KB fewer short-lived allocations and ~2,000 fewer attribute-scan iterations per run.

No local SDK is available in this CI agent; build/test verification is delegated to CI.

Trade-offs

• Very slight code duplication: the "set display name → execute → set outcome" block appears in both the fast path and the slow path's else branch. The else branch is an edge case (test has DataSourceAttribute but multiple such attributes cause TryExecuteDataSourceBasedTestsAsync to skip it), so it is never exercised on common paths.
• No behavioral changes: the fast path is only reached when both DataType != ITestDataSource and IsDataDrivenTest() is false, ensuring data-driven tests are unaffected.

Test Status

Delegated to CI. Existing unit tests in TestMethodRunnerTests.cs exercise both the new fast path (non-data-driven) and the unchanged slow path (data-driven) code paths.

Reproducibility

# Build and run adapter unit tests
./build.sh -test

🤖 Automated content by GitHub Copilot. Posted via a maintainer's GitHub token, so it appears under their account — the account owner did not write or approve this content personally. Generated by the Perf Improver workflow. · 1.7K AIC · ⌖ 26.2 AIC · ⊞ 57.7K · [◷]( · ◷)

Add this agentic workflows to your repo

To install this agentic workflow, run

gh aw add githubnext/agentics/workflows/perf-improver.md@main

[Copilot]

Pull request overview

This PR optimizes MSTest adapter execution for the common non-data-driven case by avoiding redundant attribute scans and short-lived allocations in TestMethodRunner.RunTestMethodAsync.

Changes:

• Added a private IsDataDrivenTest() helper to detect DataSourceAttribute / ITestDataSource in a single cached-attribute pass.
• Introduced a non-data-driven fast path that calls ExecuteTestAsync directly and returns its TestResult[] without allocating/spreading a List<TestResult>.
• Widened GetAggregateOutcome from List<TestResult> to IReadOnlyList<TestResult> to support both arrays and lists.

Show a summary per file

File	Description
src/Adapter/MSTestAdapter.PlatformServices/Execution/TestMethodRunner.cs	Adds a non-data-driven fast path and consolidates attribute scanning to reduce allocations and iterations.

Review details

• Files reviewed: 1/1 changed files
• Comments generated: 1
• Review effort level: Low

[Amaury Levé (Evangelink)]

Note

🤖 Automated review by GitHub Copilot. Posted via a maintainer's GitHub token, so it appears under their account — the account owner did not write or approve this content personally. Generated by the Expert Code Review workflow. To request a follow-up action, reply by tagging @copilot directly.

Review Summary

Reviewed src/Adapter/MSTestAdapter.PlatformServices/Execution/TestMethodRunner.cs against all 22 applicable dimensions. The optimization is well-reasoned and algorithmically sound for the common case, but there is one correctness concern worth addressing before merge.

---

Verdict Table

#	Dimension	Status	Notes
1	Algorithmic Correctness	⚠️SEE INLINE	Fast path bypasses results.Count == 0 guard (detail below)
2	Threading & Concurrency	✅ LGTM	No new shared-state access; all reads are already synchronized by callers
3	Security & IPC Contract Safety	✅ LGTM	No security-relevant code touched
4	Public API & Binary Compatibility	✅ LGTM	All changes are private; IReadOnlyList<T> widening is private-only
5	Performance & Allocations	✅ LGTM	Correctly eliminates List<TestResult> + spread allocation on the hot path
6	Cross-TFM Compatibility	✅ LGTM	No TFM-conditional APIs introduced
7	Resource & IDisposable Management	✅ LGTM	No disposables affected
8	Defensive Coding at Boundaries	⚠️ Same as #1	See inline comment on line 134
9	Localization & Resources	✅ LGTM	No new user-facing strings added
10	Test Isolation	✅ LGTM	No static mutation introduced
11	Assertion Quality	✅ LGTM	Tests use AwesomeAssertions (project convention)
12	Flakiness Patterns	✅ LGTM	No timing/ordering sensitivity introduced
13	Test Completeness & Coverage	✅ LGTM	Existing non-data-driven tests exercise the fast path; data-driven tests cover the slow path
14	Data-Driven Test Coverage	N/A	No new data-driven tests
15	Code Structure & Simplification	✅ LGTM	Duplication acknowledged in PR; else branch is an edge case only
16	Naming & Conventions	✅ LGTM	IsDataDrivenTest() name is clear and consistent with codebase style
17	Documentation Accuracy	✅ LGTM	XML doc and inline comments are accurate
18	Analyzer & Code Fix Quality	N/A	No src/Analyzers/ changes
19	IPC Wire Compatibility	N/A	No serialization code changed
20	Build Infrastructure & Dependencies	✅ LGTM	No build changes
21	Scope & PR Discipline	✅ LGTM	Focused, single-file, single-concern change
22	PowerShell Scripting Hygiene	N/A	No .ps1 files changed

---

The One Correctness Issue (inline comment, line 134)

DynamicDataType has only two values (None, ITestDataSource), and IsDataDrivenTest() correctly mirrors the exact attribute checks performed by TryExecuteDataSourceBasedTestsAsync (looks for DataSourceAttribute) and TryExecuteFoldedDataDrivenTestsAsync (looks for UTF.ITestDataSource). The fast-path gate is logically equivalent to the old else branch and the algorithm is correct.

The single gap: the slow path has a defensive guard at lines 190-199 that ensures RunTestMethodAsync never returns [] — because ExecuteAsync's finally block unconditionally dereferences result![0], an empty return would throw IndexOutOfRangeException rather than surfacing a clean error. The fast path omits this guard.

For all practical purposes (standard TestMethodAttribute always returns ≥ 1 result) this is benign. But a custom executor returning TestResult[0]{} would now crash on the fast path where it previously received a graceful error result. The inline comment explains the fix options.

No other issues found across the remaining 20 applicable dimensions.

[Amaury Levé (Evangelink)]

Goal and Rationale

For the vast majority of test executions — non-data-driven tests — RunTestMethodAsync was paying three unnecessary costs on every call:

1. Two attribute scans: TryExecuteDataSourceBasedTestsAsync scans all cached method attributes for DataSourceAttribute; if that returns false, TryExecuteFoldedDataDrivenTestsAsync scans them again for ITestDataSource implementors.
2. List<TestResult> allocation: a List<TestResult> is created even though only ExecuteTestAsync will be called (which already returns TestResult[]).
3. Spread allocation: return [.. results] converts the list back to an array, allocating a second short-lived TestResult[].

In a typical suite with 1,000 non-data-driven tests, this adds ~80 KB of short-lived allocations and ~2,000 extra attribute-scan iterations per run.

Approach

New IsDataDrivenTest() private method: one pass over GetCustomAttributesCached that checks attribute is DataSourceAttribute or UTF.ITestDataSource — combining both scans into one.

Fast path (new): if _test.DataType != DynamicDataType.ITestDataSourceandIsDataDrivenTest() returns false, the method:

• Calls ExecuteTestAsync directly
• Returns its TestResult[] without allocating a List<TestResult>
• Performs one attribute-cache scan instead of two

Slow path (unchanged): all data-driven tests continue through the existing List<TestResult> path.

GetAggregateOutcome signature: widened from List<TestResult> to IReadOnlyList<TestResult> so both the fast-path TestResult[] and slow-path List<TestResult> can call it (arrays and lists both implement IReadOnlyList<T>).

Performance Evidence

Per non-data-driven test, the change eliminates approximately:

• 1 List<TestResult> object (~56 bytes: 24-byte header + internal capacity-4 backing array)
• 1 TestResult[] spread allocation (~24 bytes, length 1)
• 1 redundant attribute-cache enumeration (second TryExecute... scan avoided)

For a 1,000-test non-data-driven suite: ~80 KB fewer short-lived allocations and ~2,000 fewer attribute-scan iterations per run.

No local SDK is available in this CI agent; build/test verification is delegated to CI.

Trade-offs

• Very slight code duplication: the "set display name → execute → set outcome" block appears in both the fast path and the slow path's else branch. The else branch is an edge case (test has DataSourceAttribute but multiple such attributes cause TryExecuteDataSourceBasedTestsAsync to skip it), so it is never exercised on common paths.
• No behavioral changes: the fast path is only reached when both DataType != ITestDataSource and IsDataDrivenTest() is false, ensuring data-driven tests are unaffected.

Test Status

Delegated to CI. Existing unit tests in TestMethodRunnerTests.cs exercise both the new fast path (non-data-driven) and the unchanged slow path (data-driven) code paths.

Reproducibility

# Build and run adapter unit tests
./build.sh -test

🤖 Automated content by GitHub Copilot. Posted via a maintainer's GitHub token, so it appears under their account — the account owner did not write or approve this content personally. Generated by the Perf Improver workflow. · 1.7K AIC · ⌖ 26.2 AIC · ⊞ 57.7K · [◷]( · ◷)

Add this agentic workflows to your repo

To install this agentic workflow, run

gh aw add githubnext/agentics/workflows/perf-improver.md@main

[Copilot]

Pull request overview

This PR optimizes MSTest adapter execution for the common non-data-driven case by avoiding redundant attribute scans and short-lived allocations in TestMethodRunner.RunTestMethodAsync.

Changes:

• Added a private IsDataDrivenTest() helper to detect DataSourceAttribute / ITestDataSource in a single cached-attribute pass.
• Introduced a non-data-driven fast path that calls ExecuteTestAsync directly and returns its TestResult[] without allocating/spreading a List<TestResult>.
• Widened GetAggregateOutcome from List<TestResult> to IReadOnlyList<TestResult> to support both arrays and lists.

Show a summary per file

File	Description
src/Adapter/MSTestAdapter.PlatformServices/Execution/TestMethodRunner.cs	Adds a non-data-driven fast path and consolidates attribute scanning to reduce allocations and iterations.

Review details

• Files reviewed: 1/1 changed files
• Comments generated: 1
• Review effort level: Low

[Amaury Levé (Evangelink)]

Note

🤖 Automated review by GitHub Copilot. Posted via a maintainer's GitHub token, so it appears under their account — the account owner did not write or approve this content personally. Generated by the Expert Code Review workflow. To request a follow-up action, reply by tagging @copilot directly.

Review Summary

Reviewed src/Adapter/MSTestAdapter.PlatformServices/Execution/TestMethodRunner.cs against all 22 applicable dimensions. The optimization is well-reasoned and algorithmically sound for the common case, but there is one correctness concern worth addressing before merge.

---

Verdict Table

#	Dimension	Status	Notes
1	Algorithmic Correctness	⚠️SEE INLINE	Fast path bypasses results.Count == 0 guard (detail below)
2	Threading & Concurrency	✅ LGTM	No new shared-state access; all reads are already synchronized by callers
3	Security & IPC Contract Safety	✅ LGTM	No security-relevant code touched
4	Public API & Binary Compatibility	✅ LGTM	All changes are private; IReadOnlyList<T> widening is private-only
5	Performance & Allocations	✅ LGTM	Correctly eliminates List<TestResult> + spread allocation on the hot path
6	Cross-TFM Compatibility	✅ LGTM	No TFM-conditional APIs introduced
7	Resource & IDisposable Management	✅ LGTM	No disposables affected
8	Defensive Coding at Boundaries	⚠️ Same as #1	See inline comment on line 134
9	Localization & Resources	✅ LGTM	No new user-facing strings added
10	Test Isolation	✅ LGTM	No static mutation introduced
11	Assertion Quality	✅ LGTM	Tests use AwesomeAssertions (project convention)
12	Flakiness Patterns	✅ LGTM	No timing/ordering sensitivity introduced
13	Test Completeness & Coverage	✅ LGTM	Existing non-data-driven tests exercise the fast path; data-driven tests cover the slow path
14	Data-Driven Test Coverage	N/A	No new data-driven tests
15	Code Structure & Simplification	✅ LGTM	Duplication acknowledged in PR; else branch is an edge case only
16	Naming & Conventions	✅ LGTM	IsDataDrivenTest() name is clear and consistent with codebase style
17	Documentation Accuracy	✅ LGTM	XML doc and inline comments are accurate
18	Analyzer & Code Fix Quality	N/A	No src/Analyzers/ changes
19	IPC Wire Compatibility	N/A	No serialization code changed
20	Build Infrastructure & Dependencies	✅ LGTM	No build changes
21	Scope & PR Discipline	✅ LGTM	Focused, single-file, single-concern change
22	PowerShell Scripting Hygiene	N/A	No .ps1 files changed

---

The One Correctness Issue (inline comment, line 134)

DynamicDataType has only two values (None, ITestDataSource), and IsDataDrivenTest() correctly mirrors the exact attribute checks performed by TryExecuteDataSourceBasedTestsAsync (looks for DataSourceAttribute) and TryExecuteFoldedDataDrivenTestsAsync (looks for UTF.ITestDataSource). The fast-path gate is logically equivalent to the old else branch and the algorithm is correct.

The single gap: the slow path has a defensive guard at lines 190-199 that ensures RunTestMethodAsync never returns [] — because ExecuteAsync's finally block unconditionally dereferences result![0], an empty return would throw IndexOutOfRangeException rather than surfacing a clean error. The fast path omits this guard.

For all practical purposes (standard TestMethodAttribute always returns ≥ 1 result) this is benign. But a custom executor returning TestResult[0]{} would now crash on the fast path where it previously received a graceful error result. The inline comment explains the fix options.

No other issues found across the remaining 20 applicable dimensions.

[Amaury Levé (Evangelink)]

Goal and Rationale

For the vast majority of test executions — non-data-driven tests — RunTestMethodAsync was paying three unnecessary costs on every call:

1. Two attribute scans: TryExecuteDataSourceBasedTestsAsync scans all cached method attributes for DataSourceAttribute; if that returns false, TryExecuteFoldedDataDrivenTestsAsync scans them again for ITestDataSource implementors.
2. List<TestResult> allocation: a List<TestResult> is created even though only ExecuteTestAsync will be called (which already returns TestResult[]).
3. Spread allocation: return [.. results] converts the list back to an array, allocating a second short-lived TestResult[].

In a typical suite with 1,000 non-data-driven tests, this adds ~80 KB of short-lived allocations and ~2,000 extra attribute-scan iterations per run.

Approach

New IsDataDrivenTest() private method: one pass over GetCustomAttributesCached that checks attribute is DataSourceAttribute or UTF.ITestDataSource — combining both scans into one.

Fast path (new): if _test.DataType != DynamicDataType.ITestDataSourceandIsDataDrivenTest() returns false, the method:

• Calls ExecuteTestAsync directly
• Returns its TestResult[] without allocating a List<TestResult>
• Performs one attribute-cache scan instead of two

Slow path (unchanged): all data-driven tests continue through the existing List<TestResult> path.

GetAggregateOutcome signature: widened from List<TestResult> to IReadOnlyList<TestResult> so both the fast-path TestResult[] and slow-path List<TestResult> can call it (arrays and lists both implement IReadOnlyList<T>).

Performance Evidence

Per non-data-driven test, the change eliminates approximately:

• 1 List<TestResult> object (~56 bytes: 24-byte header + internal capacity-4 backing array)
• 1 TestResult[] spread allocation (~24 bytes, length 1)
• 1 redundant attribute-cache enumeration (second TryExecute... scan avoided)

For a 1,000-test non-data-driven suite: ~80 KB fewer short-lived allocations and ~2,000 fewer attribute-scan iterations per run.

No local SDK is available in this CI agent; build/test verification is delegated to CI.

Trade-offs

• Very slight code duplication: the "set display name → execute → set outcome" block appears in both the fast path and the slow path's else branch. The else branch is an edge case (test has DataSourceAttribute but multiple such attributes cause TryExecuteDataSourceBasedTestsAsync to skip it), so it is never exercised on common paths.
• No behavioral changes: the fast path is only reached when both DataType != ITestDataSource and IsDataDrivenTest() is false, ensuring data-driven tests are unaffected.

Test Status

Delegated to CI. Existing unit tests in TestMethodRunnerTests.cs exercise both the new fast path (non-data-driven) and the unchanged slow path (data-driven) code paths.

Reproducibility

# Build and run adapter unit tests
./build.sh -test

🤖 Automated content by GitHub Copilot. Posted via a maintainer's GitHub token, so it appears under their account — the account owner did not write or approve this content personally. Generated by the Perf Improver workflow. · 1.7K AIC · ⌖ 26.2 AIC · ⊞ 57.7K · [◷]( · ◷)

Add this agentic workflows to your repo

To install this agentic workflow, run

gh aw add githubnext/agentics/workflows/perf-improver.md@main

[Copilot]

Pull request overview

This PR optimizes MSTest adapter execution for the common non-data-driven case by avoiding redundant attribute scans and short-lived allocations in TestMethodRunner.RunTestMethodAsync.

Changes:

• Added a private IsDataDrivenTest() helper to detect DataSourceAttribute / ITestDataSource in a single cached-attribute pass.
• Introduced a non-data-driven fast path that calls ExecuteTestAsync directly and returns its TestResult[] without allocating/spreading a List<TestResult>.
• Widened GetAggregateOutcome from List<TestResult> to IReadOnlyList<TestResult> to support both arrays and lists.

Show a summary per file

File	Description
src/Adapter/MSTestAdapter.PlatformServices/Execution/TestMethodRunner.cs	Adds a non-data-driven fast path and consolidates attribute scanning to reduce allocations and iterations.

Review details

• Files reviewed: 1/1 changed files
• Comments generated: 1
• Review effort level: Low

[Amaury Levé (Evangelink)]

Note

🤖 Automated review by GitHub Copilot. Posted via a maintainer's GitHub token, so it appears under their account — the account owner did not write or approve this content personally. Generated by the Expert Code Review workflow. To request a follow-up action, reply by tagging @copilot directly.

Review Summary

Reviewed src/Adapter/MSTestAdapter.PlatformServices/Execution/TestMethodRunner.cs against all 22 applicable dimensions. The optimization is well-reasoned and algorithmically sound for the common case, but there is one correctness concern worth addressing before merge.

---

Verdict Table

#	Dimension	Status	Notes
1	Algorithmic Correctness	⚠️SEE INLINE	Fast path bypasses results.Count == 0 guard (detail below)
2	Threading & Concurrency	✅ LGTM	No new shared-state access; all reads are already synchronized by callers
3	Security & IPC Contract Safety	✅ LGTM	No security-relevant code touched
4	Public API & Binary Compatibility	✅ LGTM	All changes are private; IReadOnlyList<T> widening is private-only
5	Performance & Allocations	✅ LGTM	Correctly eliminates List<TestResult> + spread allocation on the hot path
6	Cross-TFM Compatibility	✅ LGTM	No TFM-conditional APIs introduced
7	Resource & IDisposable Management	✅ LGTM	No disposables affected
8	Defensive Coding at Boundaries	⚠️ Same as #1	See inline comment on line 134
9	Localization & Resources	✅ LGTM	No new user-facing strings added
10	Test Isolation	✅ LGTM	No static mutation introduced
11	Assertion Quality	✅ LGTM	Tests use AwesomeAssertions (project convention)
12	Flakiness Patterns	✅ LGTM	No timing/ordering sensitivity introduced
13	Test Completeness & Coverage	✅ LGTM	Existing non-data-driven tests exercise the fast path; data-driven tests cover the slow path
14	Data-Driven Test Coverage	N/A	No new data-driven tests
15	Code Structure & Simplification	✅ LGTM	Duplication acknowledged in PR; else branch is an edge case only
16	Naming & Conventions	✅ LGTM	IsDataDrivenTest() name is clear and consistent with codebase style
17	Documentation Accuracy	✅ LGTM	XML doc and inline comments are accurate
18	Analyzer & Code Fix Quality	N/A	No src/Analyzers/ changes
19	IPC Wire Compatibility	N/A	No serialization code changed
20	Build Infrastructure & Dependencies	✅ LGTM	No build changes
21	Scope & PR Discipline	✅ LGTM	Focused, single-file, single-concern change
22	PowerShell Scripting Hygiene	N/A	No .ps1 files changed

---

The One Correctness Issue (inline comment, line 134)

DynamicDataType has only two values (None, ITestDataSource), and IsDataDrivenTest() correctly mirrors the exact attribute checks performed by TryExecuteDataSourceBasedTestsAsync (looks for DataSourceAttribute) and TryExecuteFoldedDataDrivenTestsAsync (looks for UTF.ITestDataSource). The fast-path gate is logically equivalent to the old else branch and the algorithm is correct.

The single gap: the slow path has a defensive guard at lines 190-199 that ensures RunTestMethodAsync never returns [] — because ExecuteAsync's finally block unconditionally dereferences result![0], an empty return would throw IndexOutOfRangeException rather than surfacing a clean error. The fast path omits this guard.

For all practical purposes (standard TestMethodAttribute always returns ≥ 1 result) this is benign. But a custom executor returning TestResult[0]{} would now crash on the fast path where it previously received a graceful error result. The inline comment explains the fix options.

No other issues found across the remaining 20 applicable dimensions.

[Amaury Levé (Evangelink)]

Goal and Rationale

For the vast majority of test executions — non-data-driven tests — RunTestMethodAsync was paying three unnecessary costs on every call:

1. Two attribute scans: TryExecuteDataSourceBasedTestsAsync scans all cached method attributes for DataSourceAttribute; if that returns false, TryExecuteFoldedDataDrivenTestsAsync scans them again for ITestDataSource implementors.
2. List<TestResult> allocation: a List<TestResult> is created even though only ExecuteTestAsync will be called (which already returns TestResult[]).
3. Spread allocation: return [.. results] converts the list back to an array, allocating a second short-lived TestResult[].

In a typical suite with 1,000 non-data-driven tests, this adds ~80 KB of short-lived allocations and ~2,000 extra attribute-scan iterations per run.

Approach

New IsDataDrivenTest() private method: one pass over GetCustomAttributesCached that checks attribute is DataSourceAttribute or UTF.ITestDataSource — combining both scans into one.

Fast path (new): if _test.DataType != DynamicDataType.ITestDataSourceandIsDataDrivenTest() returns false, the method:

• Calls ExecuteTestAsync directly
• Returns its TestResult[] without allocating a List<TestResult>
• Performs one attribute-cache scan instead of two

Slow path (unchanged): all data-driven tests continue through the existing List<TestResult> path.

GetAggregateOutcome signature: widened from List<TestResult> to IReadOnlyList<TestResult> so both the fast-path TestResult[] and slow-path List<TestResult> can call it (arrays and lists both implement IReadOnlyList<T>).

Performance Evidence

Per non-data-driven test, the change eliminates approximately:

• 1 List<TestResult> object (~56 bytes: 24-byte header + internal capacity-4 backing array)
• 1 TestResult[] spread allocation (~24 bytes, length 1)
• 1 redundant attribute-cache enumeration (second TryExecute... scan avoided)

For a 1,000-test non-data-driven suite: ~80 KB fewer short-lived allocations and ~2,000 fewer attribute-scan iterations per run.

No local SDK is available in this CI agent; build/test verification is delegated to CI.

Trade-offs

• Very slight code duplication: the "set display name → execute → set outcome" block appears in both the fast path and the slow path's else branch. The else branch is an edge case (test has DataSourceAttribute but multiple such attributes cause TryExecuteDataSourceBasedTestsAsync to skip it), so it is never exercised on common paths.
• No behavioral changes: the fast path is only reached when both DataType != ITestDataSource and IsDataDrivenTest() is false, ensuring data-driven tests are unaffected.

Test Status

Delegated to CI. Existing unit tests in TestMethodRunnerTests.cs exercise both the new fast path (non-data-driven) and the unchanged slow path (data-driven) code paths.

Reproducibility

# Build and run adapter unit tests
./build.sh -test

🤖 Automated content by GitHub Copilot. Posted via a maintainer's GitHub token, so it appears under their account — the account owner did not write or approve this content personally. Generated by the Perf Improver workflow. · 1.7K AIC · ⌖ 26.2 AIC · ⊞ 57.7K · [◷]( · ◷)

Add this agentic workflows to your repo

To install this agentic workflow, run

gh aw add githubnext/agentics/workflows/perf-improver.md@main

[Copilot]

Pull request overview

This PR optimizes MSTest adapter execution for the common non-data-driven case by avoiding redundant attribute scans and short-lived allocations in TestMethodRunner.RunTestMethodAsync.

Changes:

• Added a private IsDataDrivenTest() helper to detect DataSourceAttribute / ITestDataSource in a single cached-attribute pass.
• Introduced a non-data-driven fast path that calls ExecuteTestAsync directly and returns its TestResult[] without allocating/spreading a List<TestResult>.
• Widened GetAggregateOutcome from List<TestResult> to IReadOnlyList<TestResult> to support both arrays and lists.

Show a summary per file

File	Description
src/Adapter/MSTestAdapter.PlatformServices/Execution/TestMethodRunner.cs	Adds a non-data-driven fast path and consolidates attribute scanning to reduce allocations and iterations.

Review details

• Files reviewed: 1/1 changed files
• Comments generated: 1
• Review effort level: Low

[Amaury Levé (Evangelink)]

Note

🤖 Automated review by GitHub Copilot. Posted via a maintainer's GitHub token, so it appears under their account — the account owner did not write or approve this content personally. Generated by the Expert Code Review workflow. To request a follow-up action, reply by tagging @copilot directly.

Review Summary

Reviewed src/Adapter/MSTestAdapter.PlatformServices/Execution/TestMethodRunner.cs against all 22 applicable dimensions. The optimization is well-reasoned and algorithmically sound for the common case, but there is one correctness concern worth addressing before merge.

---

Verdict Table

#	Dimension	Status	Notes
1	Algorithmic Correctness	⚠️SEE INLINE	Fast path bypasses results.Count == 0 guard (detail below)
2	Threading & Concurrency	✅ LGTM	No new shared-state access; all reads are already synchronized by callers
3	Security & IPC Contract Safety	✅ LGTM	No security-relevant code touched
4	Public API & Binary Compatibility	✅ LGTM	All changes are private; IReadOnlyList<T> widening is private-only
5	Performance & Allocations	✅ LGTM	Correctly eliminates List<TestResult> + spread allocation on the hot path
6	Cross-TFM Compatibility	✅ LGTM	No TFM-conditional APIs introduced
7	Resource & IDisposable Management	✅ LGTM	No disposables affected
8	Defensive Coding at Boundaries	⚠️ Same as #1	See inline comment on line 134
9	Localization & Resources	✅ LGTM	No new user-facing strings added
10	Test Isolation	✅ LGTM	No static mutation introduced
11	Assertion Quality	✅ LGTM	Tests use AwesomeAssertions (project convention)
12	Flakiness Patterns	✅ LGTM	No timing/ordering sensitivity introduced
13	Test Completeness & Coverage	✅ LGTM	Existing non-data-driven tests exercise the fast path; data-driven tests cover the slow path
14	Data-Driven Test Coverage	N/A	No new data-driven tests
15	Code Structure & Simplification	✅ LGTM	Duplication acknowledged in PR; else branch is an edge case only
16	Naming & Conventions	✅ LGTM	IsDataDrivenTest() name is clear and consistent with codebase style
17	Documentation Accuracy	✅ LGTM	XML doc and inline comments are accurate
18	Analyzer & Code Fix Quality	N/A	No src/Analyzers/ changes
19	IPC Wire Compatibility	N/A	No serialization code changed
20	Build Infrastructure & Dependencies	✅ LGTM	No build changes
21	Scope & PR Discipline	✅ LGTM	Focused, single-file, single-concern change
22	PowerShell Scripting Hygiene	N/A	No .ps1 files changed

---

The One Correctness Issue (inline comment, line 134)

DynamicDataType has only two values (None, ITestDataSource), and IsDataDrivenTest() correctly mirrors the exact attribute checks performed by TryExecuteDataSourceBasedTestsAsync (looks for DataSourceAttribute) and TryExecuteFoldedDataDrivenTestsAsync (looks for UTF.ITestDataSource). The fast-path gate is logically equivalent to the old else branch and the algorithm is correct.

The single gap: the slow path has a defensive guard at lines 190-199 that ensures RunTestMethodAsync never returns [] — because ExecuteAsync's finally block unconditionally dereferences result![0], an empty return would throw IndexOutOfRangeException rather than surfacing a clean error. The fast path omits this guard.

For all practical purposes (standard TestMethodAttribute always returns ≥ 1 result) this is benign. But a custom executor returning TestResult[0]{} would now crash on the fast path where it previously received a graceful error result. The inline comment explains the fix options.

No other issues found across the remaining 20 applicable dimensions.

[Amaury Levé (Evangelink)]

Goal and Rationale

For the vast majority of test executions — non-data-driven tests — RunTestMethodAsync was paying three unnecessary costs on every call:

1. Two attribute scans: TryExecuteDataSourceBasedTestsAsync scans all cached method attributes for DataSourceAttribute; if that returns false, TryExecuteFoldedDataDrivenTestsAsync scans them again for ITestDataSource implementors.
2. List<TestResult> allocation: a List<TestResult> is created even though only ExecuteTestAsync will be called (which already returns TestResult[]).
3. Spread allocation: return [.. results] converts the list back to an array, allocating a second short-lived TestResult[].

In a typical suite with 1,000 non-data-driven tests, this adds ~80 KB of short-lived allocations and ~2,000 extra attribute-scan iterations per run.

Approach

New IsDataDrivenTest() private method: one pass over GetCustomAttributesCached that checks attribute is DataSourceAttribute or UTF.ITestDataSource — combining both scans into one.

Fast path (new): if _test.DataType != DynamicDataType.ITestDataSourceandIsDataDrivenTest() returns false, the method:

• Calls ExecuteTestAsync directly
• Returns its TestResult[] without allocating a List<TestResult>
• Performs one attribute-cache scan instead of two

Slow path (unchanged): all data-driven tests continue through the existing List<TestResult> path.

GetAggregateOutcome signature: widened from List<TestResult> to IReadOnlyList<TestResult> so both the fast-path TestResult[] and slow-path List<TestResult> can call it (arrays and lists both implement IReadOnlyList<T>).

Performance Evidence

Per non-data-driven test, the change eliminates approximately:

• 1 List<TestResult> object (~56 bytes: 24-byte header + internal capacity-4 backing array)
• 1 TestResult[] spread allocation (~24 bytes, length 1)
• 1 redundant attribute-cache enumeration (second TryExecute... scan avoided)

For a 1,000-test non-data-driven suite: ~80 KB fewer short-lived allocations and ~2,000 fewer attribute-scan iterations per run.

No local SDK is available in this CI agent; build/test verification is delegated to CI.

Trade-offs

• Very slight code duplication: the "set display name → execute → set outcome" block appears in both the fast path and the slow path's else branch. The else branch is an edge case (test has DataSourceAttribute but multiple such attributes cause TryExecuteDataSourceBasedTestsAsync to skip it), so it is never exercised on common paths.
• No behavioral changes: the fast path is only reached when both DataType != ITestDataSource and IsDataDrivenTest() is false, ensuring data-driven tests are unaffected.

Test Status

Delegated to CI. Existing unit tests in TestMethodRunnerTests.cs exercise both the new fast path (non-data-driven) and the unchanged slow path (data-driven) code paths.

Reproducibility

# Build and run adapter unit tests
./build.sh -test

🤖 Automated content by GitHub Copilot. Posted via a maintainer's GitHub token, so it appears under their account — the account owner did not write or approve this content personally. Generated by the Perf Improver workflow. · 1.7K AIC · ⌖ 26.2 AIC · ⊞ 57.7K · [◷]( · ◷)

Add this agentic workflows to your repo

To install this agentic workflow, run

gh aw add githubnext/agentics/workflows/perf-improver.md@main

[Copilot]

Pull request overview

This PR optimizes MSTest adapter execution for the common non-data-driven case by avoiding redundant attribute scans and short-lived allocations in TestMethodRunner.RunTestMethodAsync.

Changes:

• Added a private IsDataDrivenTest() helper to detect DataSourceAttribute / ITestDataSource in a single cached-attribute pass.
• Introduced a non-data-driven fast path that calls ExecuteTestAsync directly and returns its TestResult[] without allocating/spreading a List<TestResult>.
• Widened GetAggregateOutcome from List<TestResult> to IReadOnlyList<TestResult> to support both arrays and lists.

Show a summary per file

File	Description
src/Adapter/MSTestAdapter.PlatformServices/Execution/TestMethodRunner.cs	Adds a non-data-driven fast path and consolidates attribute scanning to reduce allocations and iterations.

Review details

• Files reviewed: 1/1 changed files
• Comments generated: 1
• Review effort level: Low

[Amaury Levé (Evangelink)]

Note

🤖 Automated review by GitHub Copilot. Posted via a maintainer's GitHub token, so it appears under their account — the account owner did not write or approve this content personally. Generated by the Expert Code Review workflow. To request a follow-up action, reply by tagging @copilot directly.

Review Summary

Reviewed src/Adapter/MSTestAdapter.PlatformServices/Execution/TestMethodRunner.cs against all 22 applicable dimensions. The optimization is well-reasoned and algorithmically sound for the common case, but there is one correctness concern worth addressing before merge.

---

Verdict Table

#	Dimension	Status	Notes
1	Algorithmic Correctness	⚠️SEE INLINE	Fast path bypasses results.Count == 0 guard (detail below)
2	Threading & Concurrency	✅ LGTM	No new shared-state access; all reads are already synchronized by callers
3	Security & IPC Contract Safety	✅ LGTM	No security-relevant code touched
4	Public API & Binary Compatibility	✅ LGTM	All changes are private; IReadOnlyList<T> widening is private-only
5	Performance & Allocations	✅ LGTM	Correctly eliminates List<TestResult> + spread allocation on the hot path
6	Cross-TFM Compatibility	✅ LGTM	No TFM-conditional APIs introduced
7	Resource & IDisposable Management	✅ LGTM	No disposables affected
8	Defensive Coding at Boundaries	⚠️ Same as #1	See inline comment on line 134
9	Localization & Resources	✅ LGTM	No new user-facing strings added
10	Test Isolation	✅ LGTM	No static mutation introduced
11	Assertion Quality	✅ LGTM	Tests use AwesomeAssertions (project convention)
12	Flakiness Patterns	✅ LGTM	No timing/ordering sensitivity introduced
13	Test Completeness & Coverage	✅ LGTM	Existing non-data-driven tests exercise the fast path; data-driven tests cover the slow path
14	Data-Driven Test Coverage	N/A	No new data-driven tests
15	Code Structure & Simplification	✅ LGTM	Duplication acknowledged in PR; else branch is an edge case only
16	Naming & Conventions	✅ LGTM	IsDataDrivenTest() name is clear and consistent with codebase style
17	Documentation Accuracy	✅ LGTM	XML doc and inline comments are accurate
18	Analyzer & Code Fix Quality	N/A	No src/Analyzers/ changes
19	IPC Wire Compatibility	N/A	No serialization code changed
20	Build Infrastructure & Dependencies	✅ LGTM	No build changes
21	Scope & PR Discipline	✅ LGTM	Focused, single-file, single-concern change
22	PowerShell Scripting Hygiene	N/A	No .ps1 files changed

---

The One Correctness Issue (inline comment, line 134)

DynamicDataType has only two values (None, ITestDataSource), and IsDataDrivenTest() correctly mirrors the exact attribute checks performed by TryExecuteDataSourceBasedTestsAsync (looks for DataSourceAttribute) and TryExecuteFoldedDataDrivenTestsAsync (looks for UTF.ITestDataSource). The fast-path gate is logically equivalent to the old else branch and the algorithm is correct.

The single gap: the slow path has a defensive guard at lines 190-199 that ensures RunTestMethodAsync never returns [] — because ExecuteAsync's finally block unconditionally dereferences result![0], an empty return would throw IndexOutOfRangeException rather than surfacing a clean error. The fast path omits this guard.

For all practical purposes (standard TestMethodAttribute always returns ≥ 1 result) this is benign. But a custom executor returning TestResult[0]{} would now crash on the fast path where it previously received a graceful error result. The inline comment explains the fix options.

No other issues found across the remaining 20 applicable dimensions.