[sergo] Sergo Report: DELTA 43→44 bytescomparestring + SuggestedFix Import-Injection Compile-Correctness - 2026-07-10

[github-actions[bot]]

Executive Summary

Run R61 detected a registry delta 43→44 (new 44th analyzer bytescomparestring) and audited the SuggestedFix import-injection surface. The new linter is clean and its one semantic concern was already pre-filed (#44484, closed). The proven R60 vein (SuggestedFix compile-correctness) surfaced one new, distinct, high-confidence defect: writebytestring's autofix emits io.WriteString(...) without adding the io import — a non-compiling fix that its RWSF golden test masks. Filed as 1 issue. Zero open sergo issues pre-run (team closes fast).

Key Metrics

Metric	Value
Serena LSP tools	23 (stable)
Analyzer registry	44 (Δ 43→44, bytescomparestring)
SuggestedFix linters	12 — all 12 now use RunWithSuggestedFixes (#43313 parity vein exhausted)
Open sergo issues (pre-run)	0
Issues created	1 (sg61a1)
Success score	9 / 10

Strategy (50/50 split)

Cached reuse — SuggestedFix fix-body compile-correctness (proven R60 vein, score 9)

• Re-ran the R60 probe across all 12 SuggestedFix-emitting linters, focused on assignability/compile-correctness of emitted fixes.
• Confirmed all 12 now call analysistest.RunWithSuggestedFixes → the Autofix verification-parity gap: 3 of 9 SuggestedFix linters (stringsindexcontains, sprintfint, stringreplaceminusone) have zero [Content truncated due to length] #43313 verification-parity gap is fully closed; writebytestring's named-string fix (writebytestring (new 43rd linter): autofix emits non-compiling io.WriteString(w, s) for named string types + missing RunWithSugg [Content truncated due to length] #44187) landed (string(...) wrap + RWSF).
• Adapted the vein from type assignability (R60) to import completeness: which fixes introduce a package reference the file may not import.

New exploration — 44th linter audit + import-injection convention sweep

• Audited the newest linter bytescomparestring end-to-end (see below) — clean.
• Classified all 12 SuggestedFix fixes by whether they introduce a new package: only writebytestring (io), bytescomparestring (bytes), and sprintfint (strconv) do. The rest rewrite within an already-imported package or remove code.
• Compared their import handling — exposing writebytestring as the outlier.

Findings

1. (FILED sg61a1) writebytestring autofix never adds the io import → non-compiling fix

writebytestring.buildFix (writebytestring.go:199-211) emits io.WriteString(w, s) as its only TextEdit, deferring the import to gopls (comment at :200). RunWithSuggestedFixes / raw -fix apply only the listed edits — no goimports. Any file doing w.Write([]byte(s)) without importing io (the common case — *bytes.Buffer, *strings.Builder, http.ResponseWriter, *os.File all satisfy io.Writer without needing an io import) gets undefined: io. The RWSF golden masks this because testdata (writebytestring.go:5) pre-imports io. Distinct from #44187 (named-strings only). The direct sibling bytescomparestring handles the identical add-only case correctly by injecting its import.

2. (CLEAN) bytescomparestring (44th linter) — no defect beyond the pre-filed one

Full audit

• Semantics: pre-filed bytescomparestring (new 44th linter): the "allocates" premise is inaccurate under gc — string(a) == string(b) does NOT allocate [Content truncated due to length] #44484 ("allocates premise inaccurate — string(a)==string(b) does not allocate under gc") already CLOSED; message reworded to "written the long way; use bytes.Equal for clearer intent" — no false allocation claim remains.
• Detection: uses pass.TypesInfo (IsType() + Underlying() Basic/Slice/Byte), not syntactic name matching → no shadowing/alias FP/FN.
• Autofix compile-safety: bytes.Equal(a, b) accepts named []byte types (unnamed []byte param → assignable), so no named-type break like writebytestring had.
• Import injection: addBytesImportEdit correctly handles grouped / single-convert / standalone import shapes, skips when already imported, and de-dups multi-violation files via seenImportFiles.
• Parity: wires nolint + filecheck.IsTestFile; test uses RunWithSuggestedFixes.

3. (CONTEXT) sprintfint also introduces a new package but is defensible

Why not filed

sprintfint rewrites fmt.Sprintf("%d", x) → strconv.Itoa(x), introducing strconv. Unlike writebytestring it (a) documents the goimports reliance in its Doc string (sprintfint.go:22) and (b) may need to remove the now-unused fmt import — an add+remove case that a single TextEdit cannot cleanly express. writebytestring has neither the removal complexity nor the documentation, making it the clear outlier.

Generated Task

Task (sg61a1): Inject the io import in writebytestring's SuggestedFix.

• Problem: emitted io.WriteString(...) fix references an unimported package for files lacking io.
• Fix: mirror bytescomparestring.addBytesImportEdit — ideally lift it to a shared astutil.AddImportEdit(pass, pos, pkgPath, seen) used by both linters.
• Validation: new testdata file importing only "bytes" (no io) with a .golden showing the injected import; RunWithSuggestedFixes green; grouped/single/none import shapes all compile; single import added once.
• Effort: Small–moderate (~30-60 lines + one testdata pair).

Historical Context

Registry delta timeline & recent runs

• R56 40→41 stringsindexcontains · R57 41→42 httprespbodyclose+appendbytestring · R58/R59 stable 42 · R60 42→43 writebytestring · R61 43→44 bytescomparestring.
• R58 (execcommandwithoutcontext: enclosing-scope walk crosses non-go/defer FuncLit boundaries — false positives + misleading fix (same [Content truncated due to length] #43683) execcommandwithoutcontext FuncLit boundary. R59 (Linter precision (regression): the syntactic stdlib-package-match migration never landed — 5 holdouts remain, 3 CI-enforced (#40 [Content truncated due to length] #43934) syntactic stdlib-pkg-match migration (5 holdouts). R60 (writebytestring (new 43rd linter): autofix emits non-compiling io.WriteString(w, s) for named string types + missing RunWithSugg [Content truncated due to length] #44187) writebytestring named-string non-compiling fix + RWSF.
• 61 runs · avg success 8.75 · SuggestedFix compile-correctness is now a repeatedly-productive vein (R60, R61).

Recommendations & Next-Run Focus

• Now: fix sg61a1 by extracting a shared astutil.AddImportEdit so writebytestring and bytescomparestring share one audited import-injection path (prevents future copy-paste divergence).
• R62: watch for the 45th linter; re-verify sg61a1 landed; if bytescomparestring.addBytesImportEdit is lifted/reused, audit its grouped/single/none shapes for edge FPs at reuse sites; continue the SuggestedFix compile-correctness vein on any new import-introducing autofix.

References:

• §29070240874 (R61, this run)
• Issue writebytestring (new 43rd linter): autofix emits non-compiling io.WriteString(w, s) for named string types + missing RunWithSugg [Content truncated due to length] #44187 (R60 writebytestring named-strings, landed)
• Issue bytescomparestring (new 44th linter): the "allocates" premise is inaccurate under gc — string(a) == string(b) does NOT allocate [Content truncated due to length] #44484 (bytescomparestring allocates-premise, pre-filed)

Generated by 🤖 Sergo - Serena Go Expert · 266.3 AIC · ⌖ 14.3 AIC · ⊞ 5.8K · ◷

• expires on Jul 10, 2026, 9:10 PM UTC-08:00

[github-actions[bot]]

This discussion has been marked as outdated by Sergo - Serena Go Expert.

A newer discussion is available at Discussion #44866.