[sergo] Sergo Report: DELTA-43to66-registry-catchup - 2026-08-17

[github-actions[bot]]

Executive summary

R61 audit of pkg/linters/ (Go custom analyzer subsystem). Registry grew 43 → 66 analyzers since the last run (R60, 2026-07-08) — a large gap, so this run split effort between catching up on the registry delta and continuing the proven cached-strategy sweeps. Net result: 5 findings, all previously-tracked issues confirmed resolved except one genuinely new gap, 1 issue filed.

Tool & registry updates

• Serena LSP toolset: 23 tools, unchanged.
• Detector repair: the grep -c Analyzer cmd/linters/main.go registry-delta heuristic is dead — cmd/linters/main.go was refactored to multichecker.Main(linters.All()...), so the grep always returns 0. Replaced with a new detector: compare the pkg/linters/doc.go header count ("All N active analyzers") against len(allAnalyzers) in pkg/linters/registry.go. Confirmed both are in sync at 66 — no doc-drift bug currently.
• Registry delta: 43 → 66 (23 new linters landed since R60).

Strategy: 50/50 cached vs. new exploration

Cached (reuse of proven audit classes):

• scope_boundary_funclit class — re-audited goroutinemissingrecover and walkfuncerrshadow for FuncLit-boundary correctness on recover()/defer resolution. Both clean.
• SuggestedFix compile-correctness class (the class that caught sg60a1 last run) — re-audited mapclearloop and mapdeletecheck. Both clean; comment-guards already present.
• Coverage-gate position-correctness class — audited all 13 call sites of coverage.ShouldApply across the coverage-gated linters. All clean.
• Verified sg60a1 (writebytestring) landed: isExactString/buildStringExpr now correctly distinguishes named vs. exact string types before emitting the io.WriteString autofix.
• Verified the RunWithSuggestedFixes parity gap (Autofix verification-parity gap: 3 of 9 SuggestedFix linters (stringsindexcontains, sprintfint, stringreplaceminusone) have zero [Content truncated due to length] #43313) is fully closed — all 20 SuggestedFix-bearing linters now use RWSF in their tests, not plain analysistest.Run.

New exploration: read the least-audited linter in the registry — the 66th, packagelevelmutableslicemap (ADR-52920) — against its own founding motivation rather than a generic bug-class sweep. This is the source of the one new finding below.

Findings

5 findings (4 confirmed-clean/landed, 1 new gap filed as an issue)

1. sg60a1 (writebytestring) — CONFIRMED LANDED. isExactString now correctly rejects named string types before the io.WriteString autofix; testdata badNamedString case now produces compiling output.
2. RWSF parity gap Autofix verification-parity gap: 3 of 9 SuggestedFix linters (stringsindexcontains, sprintfint, stringreplaceminusone) have zero [Content truncated due to length] #43313 — FULLY CLOSED. All 20 SuggestedFix linters verified to use RunWithSuggestedFixes in their tests.
3. goroutinemissingrecover / walkfuncerrshadow — clean. Both correctly resolve scope boundaries (named-function defer targets, filepath.Walk/WalkDir callback identification via astutil.IsPkgSelector) without over- or under-attributing across FuncLit boundaries.
4. mapclearloop / mapdeletecheck — clean. Coverage-gate position and comment-guard (astutil.HasOverlappingComment) both correct; no regression of previously-fixed mapdeletecheck: autofix silently deletes comments inside the if body — no comment guard, unlike its sibling mapclearloop #46130-class bugs.
5. sg61a1 (new) — packagelevelmutableslicemap wholesale-reassignment gap. See task below.

Generated task / issue filed

sg61a1 — packagelevelmutableslicemap (pkg/linters/packagelevelmutableslicemap/packagelevelmutableslicemap.go) only detects mutation via append()-reassignment, index-assignment, or delete(). It never detects wholesale reassignment (globalSlice = nil, = make(...), = x[:0], etc.). This is not a hypothetical: the linter's own founding motivation, issue #52683 (cited in ADR-52920), describes a JS array that is both "reset" and "appended to" — only the "appended to" half is implemented. Confirmed via testdata: zero test coverage for the reassignment shape. Filed with label sergo (temporary_id aw_sg61a1).

Metrics

• Findings this run: 5
• Tasks/issues created: 1
• Self-assessed success score: 8/10 (large registry catch-up limited depth of new-linter individual audits to one; all cached-strategy sweeps came back clean, which is a good sign for prior fixes but yields fewer new findings than an average run)
• Cumulative: 61 runs, 413 findings, 110 tasks, avg score 8.74

Historical context

• R58 (2026-07-06): ctx-family FuncLit-boundary audit, sg58a1 filed as execcommandwithoutcontext: enclosing-scope walk crosses non-go/defer FuncLit boundaries — false positives + misleading fix (same [Content truncated due to length] #43683.
• R59 (2026-07-07): syntactic-stdlib-match sweep, sg59a1 filed and landed as Linter precision (regression): the syntactic stdlib-package-match migration never landed — 5 holdouts remain, 3 CI-enforced (#40 [Content truncated due to length] #43934.
• R60 (2026-07-08): writebytestring (43rd linter) audit, sg60a1 filed — confirmed landed this run.
• R61 (this run): registry delta 43→66, detector repaired, sg61a1 filed.
• Team fix velocity remains extremely high: zero open sergo-labeled issues existed pre-run (aside from one stray unrelated probe issue).

Recommendations

• Treat the registry-delta detector fix (doc.go header vs. registry.go len(allAnalyzers)) as the new standard going forward; the old main.go grep heuristic should not be relied on again.
• The 23 linters added between R60 and R61 are still largely un-audited individually — worth prioritizing over further re-sweeps of already-clean bug classes.

Next-run focus (R62)

1. Verify sg61a1 lands.
2. Individually audit a batch of the newest (post-R60) linters not yet covered, checking first for duplicates against already-filed issues (e.g. uncheckedflushreturn: node filter omits DeferStmt — defer w.Flush() (the most common discard pattern) is never flagged #49902, Enforce-readiness: 4 clean, type-resolved linters have zero production violations — uncheckedflushreturn, bytescomparestring, ni [Content truncated due to length] #50178) before filing new ones.
3. Apply the newly-added mutation_form_incomplete pattern (enumerated match-form dispatch missing a form present in the linter's own founding issue/ADR) as a lens across other state-tracking linters.

References:

• ADR-52920: packagelevelmutableslicemap linter
• Issue #52683 (founding motivation)

Generated by 🤖 Sergo - Serena Go Expert · agent · 273.6 AIC · ⌖ 2.24 AIC · ⊞ 6.2K · ◷

• expires on Aug 17, 2026, 8:15 PM UTC-08:00

[github-actions[bot]]

This discussion was automatically closed because it expired on 2026-08-18T04:15:37.964Z.

Closed by Workflow