[sergo] Sergo Report: Stale-Cache Resync + Enforce-Readiness Trio - 2026-07-25

[github-actions[bot]]

Executive summary

Run R61 opened on a 17-day-stale cache (last persisted R60 = 2026-07-08) and re-synced it against the current tree. The custom-linter registry has grown from 43 → 57 analyzers and the registration mechanism moved (cmd/linters/main.go is now just multichecker.Main(linters.All()...); the old grep -c Analyzer main.go detector returns 0). After reconciling open issues (only one open: #47704), I audited the newest autofix linters and filed one high-confidence enforce-readiness issue covering three clean checks. A tempting second finding was probed and rejected as a false positive before filing.

Tool / registry updates

Signal	R60 (cached)	R61 (now)
Serena LSP tools	23	23 (stable)
Custom analyzers	43	57
Registry source	cmd/linters/main.go	pkg/linters/registry.go → All()
New detector	grep -c Analyzer main.go	grep -c '.Analyzer,' pkg/linters/registry.go
CI-enforced set	14–25	25 (cgo.yml:1227, +wasm :1230)
Open sergo issues	0	1 (#47704)

The doc-sync vein is now fully guarded: doc_sync_test.go::TestDocSurfacesMatchRegistryAndSpecList cross-checks registry ↔ doc.go ↔ README.md ↔ spec_test.go together (the #47190 cross-cluster gap is closed). No doc-sync finding this run.

Strategy split (50/50)

• Cached reuse — enforce-readiness sweep (proven landing vein: errortypeassertion (new 38th linter) + execcommandwithoutcontext enforce-readiness: zero production violations, add both to the [Content truncated due to length] #42416, sprintfint enforce-readiness: zero production violations under the default build — add to CI #42644, Enforce-readiness: strings-contains-family linters are clean — add stringsindexcontains to CI (0 violations); stringscountcontai [Content truncated due to length] #45186, Enforce-readiness: three clean, type-resolved linters (bytesbufferstring, mapdeletecheck, ioutildeprecated) have zero production [Content truncated due to length] #45629, mapclearloop enforce-readiness: clean, type-resolved, autofix-verified, zero production violations — but its already-enforced si [Content truncated due to length] #46341, sprintfbool (new 55th linter) enforce-readiness: clean, type-resolved, RWSF-verified, alias-safe, zero production violations — a [Content truncated due to length] #46978). Applied to the newest autofix linters after grepping production for zero-violation status.
• New exploration — named-type autofix probe. Systematically checked whether the recent bare-element autofixes reproduce the writebytestring named-string bug class (writebytestring (new 43rd linter): autofix emits non-compiling io.WriteString(w, s) for named string types + missing RunWithSugg [Content truncated due to length] #44187 / stringbytesroundtrip: the []byte(string(b)) arm mislabels the defensive-copy idiom as a "redundant round-trip" — a false-positiv [Content truncated due to length] #47704). This surfaced a candidate in stringsjoinone that was then disproven.

Findings

1. Enforce-readiness trio — appendoneelement + timenowsub + stringsjoinone (filed)

All three are clean, type-resolved, RunWithSuggestedFixes-golden-verified, nolint+filecheck-wired, and have zero production violations (default and js/wasm), yet none is in CI enforcement.

• appendoneelement — append(s, []T{x}...) → append(s, x). Type-gated on types.Universe.Lookup("append"); the surrounding append re-imposes s's element type, so no type change.
• timenowsub — time.Now().Sub(t) → time.Since(t). timeNowQualifier resolves via *types.PkgName + Imported().Path()=="time"; isSafeSinceArg restricts the argument to side-effect-free forms, neutralizing the Since-reads-arg-first evaluation-order flip (timenowsub coverage gap: isSafeSinceArg rejects side-effect-free selector/index args (e.g. time.Now().Sub(s.start)) — false nega [Content truncated due to length] #46708 coverage gap already fixed).
• stringsjoinone — strings.Join([]string{s}, sep) → s. Type-resolved via astutil.IsPkgSelector; separator-drop gated on a constant separator; formatReplacementText parenthesizes non-primary elements (the stringsjoinone autofix drops precedence: bare unparenthesized element replacement emits non-compiling / behavior-changed code in [Content truncated due to length] #47467 precedence fix).

Production greps: time.Now().Sub( → 0 sites; strings.Join([]string{ and single-element append(...{x}...) → only multi-element sites (safe). Recommendation: add -appendoneelement -timenowsub -stringsjoinone to cgo.yml:1227 + :1230.

2. stringsjoinone named-type autofix — probed, REJECTED as false positive (not filed)

Hypothesis: strings.Join([]string{s}, sep) always has type string, but the autofix emits the bare element s; for a named string element (type ID string) this would drop the implicit conversion and change the expression's static / interface-dynamic type (the #44187 class).

Disproven by Go assignability: a defined string type is not assignable to string (both are named types with identical underlying — the "at least one not a named type" clause fails), so []string{id} is a compile error. The single element must already be string or an untyped string constant, both of which stay string-typed when emitted bare. No type change; the autofix is correct. Filing this would have been the exact false-positive class Sergo hunts — so it was dropped.

Generated tasks

1. Enforce appendoneelement + timenowsub + stringsjoinone → issue #47916. Small (two-line workflow edit), zero linter-code change, validation checklist included.

Metrics

Metric	Value
Analyzers scanned	57
Linters deep-audited	5 (appendoneelement, timenowsub, stringsjoinone, bytescomparestring, stringbytesroundtrip)
Findings	6 evidence points
Issues filed	1 (enforce-readiness batch)
FPs rejected pre-filing	1 (stringsjoinone named-type)
Success score	8 / 10

Historical context

61 runs, 414 cumulative findings, avg success 8.74. The autofix-correctness / named-type vein (#44187, #44653, #45037, #47467) and the enforce-readiness vein (#45629 three-linter batch, #46341, #46978) are the two most reliable landing patterns; this run leans on both. The cache had drifted 14 linters — a reminder to push_repo_memory every run.

Recommendations & next-run focus (R62)

• Re-detect registry via grep -c '.Analyzer,' pkg/linters/registry.go vs 57; watch for a 58th linter as a fresh audit target.
• Probe autofix type/behavior correctness on the non-enforced precision-flagged linters: trimleftright (trimleftright: the "repeated rune" cutset heuristic is an arbitrary discriminator that misses the most common TrimLeft/TrimRight [Content truncated due to length] #46526), bytescomparestring (bytescomparestring (new 44th linter): the "allocates" premise is inaccurate under gc — string(a) == string(b) does NOT allocate [Content truncated due to length] #44484), stringbytesroundtrip (stringbytesroundtrip: the []byte(string(b)) arm mislabels the defensive-copy idiom as a "redundant round-trip" — a false-positiv [Content truncated due to length] #47704).
• Continue enforce-readiness for other clean, zero-violation linters among the 32 currently unenforced.

References:

• §30144591129
• §30067684687

Generated by 🤖 Sergo - Serena Go Expert · age00 · 475.4 AIC · ⌖ 14 AIC · ⊞ 5.8K · ◷

• expires on Jul 25, 2026, 9:04 PM UTC-08:00

[github-actions[bot]]

This discussion was automatically closed because it expired on 2026-07-26T05:04:04.859Z.

Closed by Workflow