[repository-quality] Repository Quality: Test Isolation & Process-Global State

[github-actions[bot]]

🎯 Repository Quality Improvement Report - Test Isolation & Process-Global State

Analysis Date: 2026-10-19
Focus Area: Test Isolation & Process-Global State (concurrency hygiene)
Strategy Type: Custom
Custom Area: Yes — the repo has ~2,100 test files and a dedicated test-race CI job. Process-global mutation in tests (os.Chdir, t.Setenv) blocks t.Parallel() and makes race coverage fragile.

Executive Summary

Production concurrency code is in good shape: all 9 non-test goroutine launch sites inspected (cli/shellcheck.go, cli/docker_images.go, cli/update_check.go, console/spinner.go, etc.) use defer recover(), semaphores or WaitGroups, and a goroutinemissingrecover linter exists.

The weaker area is tests. There are ~620 os.Chdir calls in tests (~450 in pkg/cli), ~1,190 t.Setenv/os.Chdir uses overall, and 31 time.Sleep calls in tests. os.Chdir is process-wide, so those tests cannot run in parallel. In pkg/cli, only 426 of 648 test files use t.Parallel(). make test-race covers only 4 packages with -short.

Full Analysis Report

Current State Assessment

Metric	Value	Status
Non-test go statements (incl. linter testdata)	69	✅
Production goroutines inspected with recover/ctx handling	9/9	✅
Files using sync.Mutex/RWMutex (non-test)	~26	✅
Package-level mutable-looking maps (non-registry)	~81	⚠️
os.Chdir in tests	620 (cli: ~450)	❌
t.Setenv + os.Chdir in tests	1,193	⚠️
cli test files with t.Parallel()	426 / 648	⚠️
time.Sleep in tests	31	⚠️
-race coverage	4 packages, -short	⚠️
Test/Source LOC	679k / 374k	✅

Findings

Strengths

• Goroutines consistently recover panics and honour context cancellation.
• syncutil/onceloader.go provides shared lazy-init primitives.
• CI already runs test-race.

Areas for Improvement

• ❌ High: os.Chdir in tests is process-global and incompatible with t.Parallel(). Go 1.24+ offers t.Chdir, which restores the directory automatically but still forbids parallel use. Tests should take explicit directory parameters instead.
• ⚠️ Medium: ~81 package-level maps in pkg/* that may be written at runtime need an audit. Unwritten ones should be documented read-only, or guarded.
• ⚠️ Medium: 31 time.Sleep calls in tests are flake risks. Replace with channels, require.Eventually or fake clocks.
• ⚠️ Low: test-race omits pkg/agentdrain, pkg/workqueue, pkg/syncutil, pkg/console (console is included), and pkg/parser virtual FS users beyond the listed four.

---

🤖 Tasks for Copilot Agent

NOTE TO PLANNER AGENT: Split the following tasks into individual work items.

Improvement Tasks

Task 1: Replace raw os.Chdir in tests with t.Chdir

Priority: High
Estimated Effort: Large (do per-package batches)
Focus Area: Test isolation

Description: Raw os.Chdir plus manual deferred restore is error-prone and leaks cwd on failure. Migrate to t.Chdir(dir) (Go 1.24+; repo is on 1.26).

Acceptance Criteria:

• Zero os.Chdir in pkg/**/*_test.go, except where the test needs raw semantics (documented)
• make test-unit passes
• No new t.Parallel() combined with t.Chdir

Code Region: pkg/cli/*_test.go, pkg/workflow/*_test.go, pkg/parser/*_test.go

Migrate os.Chdir usages in test files to t.Chdir(dir), removing manual defer-restore blocks. Do it in batches per package (start with pkg/cli files containing the most occurrences). Run `make fmt` and `go test` on each touched package.

---

Task 2: Add a ratchet linter/test banning new os.Chdir in tests

Priority: Medium
Estimated Effort: Small
Focus Area: Test isolation

Description: Prevent regressions using the existing pkg/linters framework.

Acceptance Criteria:

• New analyzer flags os.Chdir in _test.go files, with testdata
• Registered in the linter runner and an allowlist for legacy files

Code Region: pkg/linters/

Following the pattern of existing analyzers (e.g. pkg/linters/timesleepnocontext), add a `testoschdir` analyzer that reports os.Chdir in _test.go files and suggests t.Chdir. Include testdata and register it.

---

Task 3: Remove time.Sleep from tests

Priority: Medium
Estimated Effort: Medium
Focus Area: Flake reduction

Description: 31 sleeps in tests; replace with synchronisation.

Acceptance Criteria:

• Each sleep replaced by channel/Eventually/injected clock, or justified in a comment
• Tests pass with -race -count=20 locally

Code Region: grep -rn "time.Sleep" pkg --include=*_test.go

Find time.Sleep in *_test.go under pkg/, replace with deterministic synchronisation (channels, sync.WaitGroup, assert.Eventually). Verify with `go test -race -count=10` for each package touched.

---

Task 4: Audit package-level mutable maps/slices

Priority: Medium
Estimated Effort: Medium
Focus Area: Concurrency safety

Description: Identify package-level maps written after init; guard with a mutex or convert to constructors.

Acceptance Criteria:

• Written-at-runtime globals are guarded or eliminated
• A short audit table of the rest is added to code comments or the packagelevelmutableslicemap linter allowlist

Code Region: pkg/parser/schema_*.go, pkg/workflow/*registry*.go, pkg/constants/

Run the packagelevelmutableslicemap linter and grep `^var .* = map\[` in pkg/. For each, determine whether it is mutated after init. Guard with sync.RWMutex or sync.Once when mutated; otherwise leave unchanged.

---

Task 5: Broaden make test-race

Priority: Low
Estimated Effort: Small
Focus Area: CI/CD

Description: Add concurrency-heavy packages to the race job.

Acceptance Criteria:

• agentdrain, workqueue, syncutil, logger added to test-race
• CI duration increase is under 2 minutes

Code Region: Makefile:95-98

Update the test-race target to include ./pkg/agentdrain ./pkg/workqueue ./pkg/syncutil ./pkg/logger and verify they pass.

---

📊 Historical Context

Previous Focus Areas

Date	Focus Area	Type	Custom	Key Outcomes
2026-10-13	Large Source File Decomposition	Custom	Y	4 tasks
2026-10-14	Library Robustness	Custom	Y	4 tasks
2026-10-15	Error Message Quality	Custom	Y	4 tasks
2026-10-16	Subprocess Execution Hygiene	Custom	Y	4 tasks

---

🎯 Recommendations

Immediate Actions (This Week)

1. Add the os.Chdir ratchet linter (Task 2) — Priority: High

Short-term Actions (This Month)

1. Migrate os.Chdir to t.Chdir (Task 1); remove sleeps (Task 3) — Priority: Medium

Long-term Actions (This Quarter)

1. Audit globals and widen race coverage (Tasks 4–5) — Priority: Low

---

📈 Success Metrics

• os.Chdir in tests: 620 → <50
• time.Sleep in tests: 31 → <10
• Race-tested packages: 4 → 8

---

Next Steps

1. Review and prioritise the tasks above
2. Assign tasks to Copilot coding agent via planner agent
3. Track progress on improvement items
4. Re-evaluate this focus area in 1 month

---

Generated by Repository Quality Improvement Agent
Next analysis: 2026-10-20 — Focus area selected by diversity algorithm

Generated by ⚡ Repository Quality Improvement Agent · copilot · auto · 19.4 AIC · ⌖ 0.695 AIC · ⊞ 9.8K · ◷

• expires on Oct 20, 2026, 5:07 AM UTC-08:00

[github-actions[bot]]

This discussion was automatically closed because it expired on 2026-10-20T13:07:31.013Z.

Closed by Workflow