Context
Audit in docs/v1-rc1-test-audit.md §1.
module/pkg/nccl/fr_parser measures 77.7% statements coverage on chore/v1-rc1-knowledge-gaps (target ≥ 80%). Func-level: parser_containers.go 68%, synthesize.go 69%. Both files are exercised heavily by FuzzParseFRPickle (module/pkg/nccl/fr_parser/fuzz_test.go:28), but the PR-gate fuzz window is 30 s and the nightly is 10 min — committed seeds don't reliably hit every branch on every PR.
Fix
Promote ten existing nightly-fuzz crash-then-fixed corpus entries from module/pkg/nccl/fr_parser/testdata/fuzz/FuzzParseFRPickle/ into deterministic f.Add(...) seeds in fuzz_test.go. Target ≥ 85% statements coverage on the PR-gate run (without depending on fuzz mutation).
Acceptance
cd module && GOWORK=off go test -cover ./pkg/nccl/fr_parser/ reports ≥ 85.0%.- Each promoted seed has a comment naming the original crash signature it pinned.
make ci still passes the 30 s fuzz gate.
Effort
S (a few hours). Adopt-over-build: stdlib testing.F only.
Context
Audit in
docs/v1-rc1-test-audit.md§1.module/pkg/nccl/fr_parsermeasures 77.7% statements coverage onchore/v1-rc1-knowledge-gaps(target ≥ 80%). Func-level:parser_containers.go68%,synthesize.go69%. Both files are exercised heavily byFuzzParseFRPickle(module/pkg/nccl/fr_parser/fuzz_test.go:28), but the PR-gate fuzz window is 30 s and the nightly is 10 min — committed seeds don't reliably hit every branch on every PR.Fix
Promote ten existing nightly-fuzz crash-then-fixed corpus entries from
module/pkg/nccl/fr_parser/testdata/fuzz/FuzzParseFRPickle/into deterministicf.Add(...)seeds infuzz_test.go. Target ≥ 85% statements coverage on the PR-gate run (without depending on fuzz mutation).Acceptance
cd module && GOWORK=off go test -cover ./pkg/nccl/fr_parser/reports ≥ 85.0%.make cistill passes the 30 s fuzz gate.Effort
S (a few hours). Adopt-over-build: stdlib
testing.Fonly.