Remove VSTest run/discovery context from PlatformServices (Phase 6d-2)

[Evangelink]

Phase 6d-2 of the PlatformServices platform-agnostic initiative

Removes the VSTest IRunContext/IDiscoveryContext types entirely from MSTestAdapter.PlatformServices by extracting the only two members ever read off them — .RunSettings.SettingsXml and .TestRunDirectory — at the adapter boundary and passing them in.

What changed

• DeploymentContext (the { TestRunDirectory, RunSettingsXml } DTO from 6a) is un-guarded and becomes the single execution-inputs carrier.
• Execution: MSTestExecutor builds new DeploymentContext(runContext?.TestRunDirectory, runContext?.RunSettings?.SettingsXml) and injects it into TestExecutionManager.RunTestsAsync/ExecuteTestsAsync/ExecuteTestsInSourceAsync/Deploy.
• Discovery: MSTestDiscoverer passes the run-settings XML string into UnitTestDiscoverer.DiscoverTests/DiscoverTestsInSource; MSTestDiscovererHelpers.InitializeDiscovery and MSTestSettings.PopulateSettings take string? settingsXml.

No behavior change

Only .SettingsXml and .TestRunDirectory were ever read, so this is byte-for-byte. The PopulateSettings guard change is provably equivalent (RunSettingsFileHasMSTestSettings(null) returns false). IRunContext/IDiscoveryContext are now absent from PlatformServices code (doc comments only); the remaining ObjectModel.Adapter surface is the IFrameworkHandle-backed deploy/recorder/logger handles.

Verification

Full build all TFMs; PlatformServices.UnitTests 897/935; MSTestAdapter.UnitTests 21; MSTest.IntegrationTests 47 pass/1 skip (incl. TestCaseFilteringTests + deployment); PlatformServices.Desktop.IntegrationTests 15 (AppDomain deployment). Expert-reviewer: no material findings, definitive byte-for-byte verdict.

Base / stacking

Stacked on Phase 6d-1 (#9591). Base = dev/amauryleve/vstest-decoupling-filter; review/merge after #9591.

Remaining

6e (relocate bridges + deep UnitTestElement<->TestCase + EngineConstants; close #9568/#9573), then 7 (drop the ObjectModel/ObjectModel.Utilities PackageReference + guard test).

[Copilot]

Copilot wasn't able to review this pull request because it exceeds the maximum diff size. Try reducing the number of changed files and lines, and requesting a review from Copilot again.

[github-actions]

🧪 Test quality grade — PR #9621

48 test methods graded across 10 files. Most changes are refactoring-only signature updates; five genuinely new test methods were added in TestCaseDiscoverySinkTests and UnitTestElementTests. Main quality concerns: one hard-coded Windows path, two integration tests with no explicit assertion, and several tests using SequenceEqual.Should().BeTrue() instead of the more diagnostic .Should().Equal(). The 5 new tests are all A-grade.

Grade	Test	Notes
D (60–69)	mod TestExecutionManagerTests. RunTestsForTestShouldLoadSourceFromDeploymentDirectoryIfDeployed	Hard-coded @"C:\temp" in both mock setup and Verify — fails on Linux; replace with a platform-agnostic temp path.
C (70–79)	mod DesktopTestSourceHostTests. ChildDomainResolutionPathsShouldHaveSearchDirectoriesSpecifiedInRunsettings	No explicit assertion; test only confirms AppDomainUtilities.CreateInstance does not throw — add an assertion on the returned instance.
C (70–79)	mod DesktopTestSourceHostTests. ParentDomainShouldHonorSearchDirectoriesSpecifiedInRunsettings	No explicit assertion; test only verifies Assembly.Load does not throw — add at least one assertion on the loaded assembly.
C (70–79)	mod TcmTestPropertiesProviderTests. GetTcmPropertiesShouldHandleDuplicateTestsProperlyFromTestCase	75-line body with three independent verification cycles — split into separate focused tests per call scenario.
C (70–79)	mod TestContextImplementationTests. WritesFromBackgroundThreadShouldNotThrow	No explicit assertion; test only verifies concurrent writes don't throw — add a buffer-state assertion after t.Join().
C (70–79)	mod TestExecutionManagerTests. RunTestsForTestShouldPassInDeploymentInformationAsPropertiesToTheTest	Only verifies GetProperties was called once with any argument — add an assertion on the actual properties injected into the test context.
B (80–89)	mod TcmTestPropertiesProviderTests. GetTcmPropertiesShouldCopyMultiplePropertiesCorrectlyFromTestCase	50-line body with two verification cycles; thorough assertions — consider splitting into separate tests for each call scenario.
B (80–89)	mod TcmTestPropertiesProviderTests. GetTcmPropertiesShouldReturnEmptyDictionaryIfTestCaseIsNull	Single null-check assertion — complete for this narrow boundary condition.
B (80–89)	mod TcmTestPropertiesProviderTests. GetTcmPropertiesShouldReturnEmptyDictionaryIfTestCaseIdIsZero	Single null assertion after 15-property setup — briefly asserting which property triggered the null return would improve diagnostics.
B (80–89)	mod TypeEnumeratorTests. GetTestsShouldReturnBaseTestMethodsFromAnotherAssemblyByConfiguration	NotBeNull + Contain checks cover the key case — also asserting total count would guard against unexpected extra discovered methods.
B (80–89)	mod TypeEnumeratorTests. GetTestsShouldReturnBaseTestMethodsFromAnotherAssemblyByDefault	Same structure as ByConfiguration — asserting total test count would strengthen coverage.
B (80–89)	mod UnitTestDiscovererTests. DiscoverTestsInSourceShouldSendBackTestCasesDiscovered	Times.AtLeastOnce leaves exact count unchecked — pin to Times.Exactly(N) or verify specific test FQNs.
B (80–89)	new UnitTestElementTests. WithUpdatedSourceShouldReturnSameInstanceWhenSourceUnchanged	Single identity assertion — clear and correct for this same-source edge case.
B (80–89)	mod TestExecutionManagerTests. RunTestsForASingleTestShouldSendSingleResult	SequenceEqual.Should().BeTrue() obscures which element differed on failure — prefer .Should().Equal(...).
B (80–89)	mod TestExecutionManagerTests. RunTestsForIgnoredTestShouldSendResultsMarkingIgnoredTestsAsSkipped	Asserts only index-0 elements — adding count checks would catch spurious extra events.
B (80–89)	mod TestExecutionManagerTests. RunTestsForMultipleSourcesShouldRunEachTestJustOnce	Checks count only — also verifying that no specific test ran twice would make the intent explicit.
B (80–89)	mod TestExecutionManagerTests. RunTestsForMultipleTestShouldSendMultipleResults	Mixes SequenceEqual.Should().BeTrue() with direct Should() calls — standardise on .Should().Equal().
B (80–89)	mod TestExecutionManagerTests. RunTestsForTestShouldPassInTcmPropertiesAsPropertiesToTheTest	Delegates to VerifyTcmProperties loop; thorough, but failure messages point to the helper, not the test scenario.
B (80–89)	mod TestExecutionManagerTests. RunTestsForTestShouldPassInTestRunParametersInformationAsPropertiesToTheTest	Checks only webAppUrl — also assert webAppUserName to cover both configured parameters.
B (80–89)	mod TestExecutionManagerTests. RunTestsForTestWithFilterShouldSendResultsForFilteredTests	SequenceEqual.Should().BeTrue() pattern gives poor failure diagnostics — prefer .Should().Equal(...).
A (90–100)	mod TcmTestPropertiesProviderTests. GetTcmPropertiesShouldGetAllPropertiesFromTestCase	All 15 TCM properties verified via helper loop; clear AAA structure.
A (90–100)	new TestCaseDiscoverySinkTests. SendTestElementShouldAddEachTestElementInOrder	Count + identity at both indices — fully characterises insertion order.
A (90–100)	new TestCaseDiscoverySinkTests. SendTestElementShouldAddTheTestElement	Count + identity + not-null — three distinct assertions on a single add.
A (90–100)	new TestCaseDiscoverySinkTests. TestCaseDiscoverySinkConstructorShouldInitializeTests	Checks not-null and zero count — fully characterises the initialised state.
A (90–100)	mod TestContextImplementationTests. DisplayMessageShouldForwardToIMessageLogger	Three exact mock verifies per message level — exhaustive coverage of all paths.
A (90–100)	mod TestPropertyAttributeTests. GetTestMethodInfoShouldAddPropertiesFromContainingClassAndBaseClassesAndOverriddenMethodsCorrectly_ OverriddenIsNotTestMethod	Exact name-value assertions for all six traits in order; thorough and precise.
A (90–100)	mod TestPropertyAttributeTests. GetTestMethodInfoShouldAddPropertiesFromContainingClassAndBaseClassesAndOverriddenMethodsCorrectly_ OverriddenIsTestMethod	Exact name-value assertions for all six traits in order.
A (90–100)	mod TestPropertyAttributeTests. GetTestMethodInfoShouldAddPropertiesFromContainingClassCorrectly	Exact name-value checks for all three traits at each index; clean AAA.
A (90–100)	mod UnitTestDiscovererTests. DiscoverTestsInSourceShouldNotThrowWhenTreatDiscoveryWarningsAsErrorsIsFalse	No-throw path verified; exact logger call counts for both warning and error levels checked.
A (90–100)	mod UnitTestDiscovererTests. DiscoverTestsInSourceShouldThrowOnFileNotFound	Exception type + message pattern assertion; fully specifies the expected error.
A (90–100)	mod UnitTestDiscovererTests. DiscoverTestsInSourceShouldThrowWhenTreatDiscoveryWarningsAsErrorsIsTrue	Exception assertion + two logger verifies — covers primary outcome and side-effects.
A (90–100)	mod UnitTestDiscovererTests. DiscoverTestsShouldThrowOnFileNotFound	Exception + message pattern; clean and specific.
A (90–100)	mod UnitTestDiscovererTests. SendTestCasesShouldNotSendAnyTestCasesIfFilterError	Both FQNs verified Times.Never — covers the full negative filter-error path.
A (90–100)	mod UnitTestDiscovererTests. SendTestCasesShouldNotSendAnyTestCasesIfThereAreNoTestElements	Concise negative test — verifies no test case was sent for an empty input.
A (90–100)	mod UnitTestDiscovererTests. SendTestCasesShouldSendAllTestCaseData	Each specific FQN verified Times.Once; precise mock-call assertions.
A (90–100)	mod UnitTestDiscovererTests. SendTestCasesShouldSendAllTestCasesIfGetTestCaseFilterNotPresent	Both FQNs sent once when the filter API is absent — correct boundary case.
A (90–100)	mod UnitTestDiscovererTests. SendTestCasesShouldSendAllTestCasesIfNullFilterExpression	Both FQNs sent once for null expression — clear null-filter boundary test.
A (90–100)	mod UnitTestDiscovererTests. SendTestCasesShouldSendFilteredTestCasesIfValidFilterExpression	Matching FQN Times.Once, non-matching Times.Never — clean positive/negative pair.
A (90–100)	new UnitTestElementTests. GetOrCreateHostTestCaseShouldMaterializeAndCacheWhenNoHandle	Null start-state + not-null result + same-instance cache check — verifies the full caching contract.
A (90–100)	new UnitTestElementTests. GetOrCreateHostTestCaseShouldReturnHostHandleWhenPresent	Identity check confirms the pre-set handle is returned unchanged.
A (90–100)	new UnitTestElementTests. WithUpdatedSourceShouldReturnCloneWithNewSourceAndLeaveOriginalUnchanged	Four distinct invariants verified: clone identity, new source, original immutability, separate method object.
A (90–100)	mod TestExecutionManagerTests. RunTestsForCancellationTokenCanceledSetToTrueShouldSendZeroResults	Exception assertion + three empty-list checks — fully characterises the cancelled-run contract.
A (90–100)	mod TestExecutionManagerTests. RunTestsForTestShouldCleanupAfterExecution	Verifies both load and cleanup call ordering by index; meaningful sequence assertion.
A (90–100)	mod TestExecutionManagerTests. RunTestsForTestShouldDeployBeforeExecution	Verifies deploy-before-load ordering; important behavioral constraint.
A (90–100)	mod TestExecutionManagerTests. RunTestsForTestShouldNotCleanupOnTestFailure	Times.Never Moq verify on cleanup when a test fails; precise negative assertion.
A (90–100)	mod TestExecutionManagerTests. RunTestsForTestWithFilterErrorShouldSendZeroResults	All three output lists verified empty on filter error; thorough boundary check.
A (90–100)	mod TestExecutionManagerTests. RunTestsShouldClearSessionParametersAcrossRuns	Verifies updated parameter value after second run — catches stale-session-parameter bugs.
A (90–100)	mod TestExecutionManagerTests. SendTestResults_ WhenUnitTestResultsIsEmpty_ RecordsEndWithoutResult	Equal for end-list + BeEmpty for results; clean dual-assertion boundary test.

_{This advisory comment was generated automatically. Grades are heuristic
and informational — they do not block merging. Re-run with
/grade-tests.}

🤖 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 Grade Tests on PR (on open / sync) workflow. · 371.7 AIC · ⌖ 9.87 AIC · ⊞ 9.5K · [◷]( · ◷)