Repository navigation
Description
Analysis of commit a760075 ("Consolidate reports across retry attempts", #10542)
Assignee: @copilot
Summary
The retry-attempt consolidation feature introduced the same "length-prefixed component" identity-encoding trick independently in two different report mergers: JUnitReportMerger and CtrfReportMerger. Both need a way to build a deterministic, collision-free identity key out of several nullable string components (suite name, test name, file path, etc.), and both solve it with the identical, non-obvious encoding scheme: for each component, append length ?? -1, then :, then the component's text, so that ambiguous concatenations (e.g. "A" + "BC" vs "AB" + "C") can never collide.
Duplication Details
Pattern: Length-prefixed identity component encoding
-
Severity: Medium (subtle, correctness-critical logic — a divergence between the two copies could silently reintroduce identity collisions in one merger but not the other)
-
Occurrences: 2 independent implementations of the same algorithm
-
Locations:
src/Platform/Microsoft.Testing.Extensions.JUnitReport/JUnitReportMerger.cs,BuildIdentity(params string?[] components)(~lines 353-362) and its callersBuildSuiteIdentity/BuildTestIdentity(lines 339-352)src/Platform/Microsoft.Testing.Extensions.CtrfReport/CtrfReportMerger.RetryCollapsing.cs,AppendIdentityComponent(StringBuilder identity, string? component)(~line 157) and its use insideGetTestIdentity(lines 136-153)
-
Code Sample (JUnitReportMerger.cs):
private static string BuildIdentity(params string?[] components) { var identity = new StringBuilder(); foreach (string? component in components) { identity.Append(component?.Length ?? -1) .Append(':') .Append(component); } return identity.ToString(); }
-
Code Sample (CtrfReportMerger.RetryCollapsing.cs):
private static void AppendIdentityComponent(StringBuilder identity, string? component) => identity.Append('\u001f').Append(component?.Length ?? -1).Append(':').Append(component);
Both encode "unambiguous concatenation of variable-length nullable strings" with the exact same length-prefix technique (the CTRF version additionally prefixes a \u001f separator, which is a trivial variation). The underlying algorithm — and the reasoning documented in the CTRF XML doc comment about why plain separation can collide — is identical to what JUnitReportMerger.BuildIdentity already needed to solve for suite/test identity a few files away in the same PR.
Impact Analysis
- Maintainability: Any future fix to the encoding scheme (e.g., switching separators, handling additional edge cases such as components containing the sentinel character) must be applied in two places; it is easy to update one and forget the other.
- Bug Risk: Because this is a security/correctness-sensitive dedup key (used to decide whether two test rows represent the same logical test across retry attempts), a fix applied to only one copy would silently reintroduce collisions in the other merger, and there is no shared test suite that would catch the divergence.
- Code Bloat: Small in absolute size, but the duplicated reasoning (documented at length in CTRF's XML doc comments) is real design/domain knowledge that currently lives in only one of the two copies, making the JUnit version's identical technique look uncommented and coincidental.
Refactoring Recommendations
-
Extract a shared identity-encoding helper
- Extract common functionality to:
src/Platform/Microsoft.Testing.Platform.Extensions/ArtifactPostProcessing/IdentityKeyBuilder.cs(or a similarly-scoped internal shared utility referenced by bothMicrosoft.Testing.Extensions.JUnitReportandMicrosoft.Testing.Extensions.CtrfReportprojects, e.g. via a shared internal project or linked file) - Estimated effort: ~1-2 hours (small, low-risk change; both call sites already isolate the encoding behind a private static helper, so the refactor is mostly moving the method and updating call sites/tests)
- Benefits: Single source of truth for the collision-avoidance reasoning; a future fix (e.g., handling the sentinel character appearing inside a component) needs to happen once; makes the shared design intent explicit instead of duplicated tribal knowledge.
- Extract common functionality to:
-
At minimum, cross-link the two implementations with comments
- If extracting a shared type is not desired (e.g., to keep the two extension packages fully independent), add a comment in each file pointing at the other, explaining that the encoding scheme must stay in sync, and add a unit test in one project asserting parity of behavior/documentation intent.
- Estimated effort: ~30 minutes
- Benefits: Lower risk than a full extraction, but at least prevents silent divergence.
Implementation Checklist
- Review duplication findings
- Decide between full extraction vs. documented parity (recommendation 1 vs 2)
- Create refactoring plan
- Implement changes
- Update/add unit tests covering both mergers' identity-encoding behavior
- Verify no functionality broken (existing
JUnitReportMergerTests/CtrfReportMergerTestsstill pass)
Analysis Metadata
- Analyzed Files:
JUnitArtifactPostProcessor.cs,JUnitReportMerger.cs,CtrfArtifactPostProcessor.cs,CtrfReportMerger.cs,CtrfReportMerger.RetryCollapsing.cs,CtrfReportMerger.FileMerge.cs,RetryArtifactProcessor.cs, plus other post-processors in the same commit - Detection Method: Semantic code analysis of recent commits
- Commit: a760075
- Analysis Date: 2026-08-14
🤖 Automated content by GitHub Copilot. Generated by the Duplicate Code Detector workflow. · auto · 49.2 AIC · ⌖ 2.27 AIC · ⊞ 12.9K · [◷]( · ◷)
Add this agentic workflow to your repo
To install this agentic workflow, run
gh aw add githubnext/agentics/workflows/duplicate-code-detector.md@main
- expires on Aug 16, 2026, 4:49 AM UTC
Activity
github-actions commentedon Aug 16, 2026
This issue was automatically closed because it expired on 2026-08-16T04:49:07.749Z.