test: stabilize output polling coverage (Fixes #564)

[Karthik Nadig (karthiknadig)]

Make the output tests exercise their real polling helpers deterministically, so coverage no longer depends on whether the writer finishes before the first poll.

• Coordinate the first observed pending state with bounded, one-shot channels.
• Cover both pending and initially-ready success/error paths.
• Keep production code, coverage exclusions, and regression budgets unchanged.

Validation: 53 pet-jsonrpc all-target/all-feature tests; workspace all-target/all-feature Clippy; mandatory Rust precommit checks; independent Reviewer LGTM. Signed commit 840e70a.

Fixes #564

[github-actions]

Performance Report (Linux)

Result: ✅ Within regression budgets

Metric	PR	Baseline	Delta	Change	Blocking budget	Status
Server startup P50	1ms	1ms	+0ms	+0.0%	>5ms and >100%	➖
Server startup P95	1ms	1ms	+0ms	+0.0%	>50ms and >200%	➖
Discovery duration P50	68ms	48ms	+20ms	+41.7%	>25ms and >30%	🔺
Discovery duration P95	73ms	52ms	+21ms	+40.4%	>50ms and >50%	🔺
Startup-to-first environment P50	14ms	10ms	+4ms	+40.0%	>20ms and >100%	🔺
Startup-to-first environment P95	17ms	12ms	+5ms	+41.7%	>25ms and >100%	🔺
Cold discovery duration P50	148ms	108ms	+40ms	+37.0%	>100ms and >50%	🔺
Refresh round-trip P50	68ms	48ms	+20ms	+41.7%	>25ms and >30%	🔺
Refresh round-trip P95	73ms	52ms	+21ms	+40.4%	>50ms and >50%	🔺
Request-to-first environment P50	13ms	9ms	+4ms	+44.4%	>20ms and >100%	🔺
Request-to-first environment P95	15ms	11ms	+4ms	+36.4%	>25ms and >100%	🔺
Cold refresh round-trip P50	148ms	108ms	+40ms	+37.0%	>100ms and >50%	🔺

Workload	PR	Baseline
Environments	5	5
Managers	1	1

A regression must exceed both the documented absolute and relative budget. Environment and manager inventories must match exactly within the same inventory schema.

[github-actions]

Test Coverage Report (Linux)

Result: ✅ Within regression budget

Metric	PR	Baseline	Delta
Lines	85.580%	85.548%	+0.033pp
Functions	88.400%	88.368%	+0.032pp

Allowed numerical tolerance: 0.01 percentage points.

[github-actions]

Performance Report (macOS)

Result: ✅ Within regression budgets

Metric	PR	Baseline	Delta	Change	Blocking budget	Status
Server startup P50	86ms	78ms	+8ms	+10.3%	>100ms and >50%	🔺
Server startup P95	789ms	868ms	-79ms	-9.1%	>750ms and >100%	✅
Discovery duration P50	129ms	146ms	-17ms	-11.6%	>100ms and >50%	✅
Discovery duration P95	210ms	223ms	-13ms	-5.8%	>300ms and >100%	✅
Startup-to-first environment P50	125ms	109ms	+16ms	+14.7%	>150ms and >50%	🔺
Startup-to-first environment P95	181ms	175ms	+6ms	+3.4%	>250ms and >100%	🔺
Cold discovery duration P50	319ms	269ms	+50ms	+18.6%	>250ms and >50%	🔺
Refresh round-trip P50	130ms	147ms	-17ms	-11.6%	>250ms and >50%	✅
Refresh round-trip P95	213ms	231ms	-18ms	-7.8%	>300ms and >100%	✅
Request-to-first environment P50	35ms	41ms	-6ms	-14.6%	>50ms and >50%	✅
Request-to-first environment P95	79ms	74ms	+5ms	+6.8%	>100ms and >100%	🔺
Cold refresh round-trip P50	319ms	270ms	+49ms	+18.1%	>600ms and >50%	🔺

Workload	PR	Baseline
Environments	10	10
Managers	1	1

A regression must exceed both the documented absolute and relative budget. Environment and manager inventories must match exactly within the same inventory schema.

[github-actions]

Performance Report (Windows)

Result: ✅ Within regression budgets

Metric	PR	Baseline	Delta	Change	Blocking budget	Status
Server startup P50	7ms	11ms	-4ms	-36.4%	>10ms and >50%	✅
Server startup P95	11ms	14ms	-3ms	-21.4%	>50ms and >100%	✅
Discovery duration P50	107ms	162ms	-55ms	-34.0%	>150ms and >50%	✅
Discovery duration P95	118ms	173ms	-55ms	-31.8%	>250ms and >100%	✅
Startup-to-first environment P50	18ms	29ms	-11ms	-37.9%	>25ms and >50%	✅
Startup-to-first environment P95	22ms	45ms	-23ms	-51.1%	>100ms and >100%	✅
Cold discovery duration P50	106ms	164ms	-58ms	-35.4%	>150ms and >50%	✅
Refresh round-trip P50	108ms	163ms	-55ms	-33.7%	>150ms and >50%	✅
Refresh round-trip P95	118ms	174ms	-56ms	-32.2%	>250ms and >100%	✅
Request-to-first environment P50	11ms	18ms	-7ms	-38.9%	>25ms and >50%	✅
Request-to-first environment P95	14ms	34ms	-20ms	-58.8%	>100ms and >100%	✅
Cold refresh round-trip P50	106ms	165ms	-59ms	-35.8%	>150ms and >50%	✅

Workload	PR	Baseline
Environments	8	8
Managers	1	1

A regression must exceed both the documented absolute and relative budget. Environment and manager inventories must match exactly within the same inventory schema.

[github-actions]

Test Coverage Report (Windows)

Result: ✅ Within regression budget

Metric	PR	Baseline	Delta
Lines	83.130%	83.075%	+0.055pp
Functions	85.791%	85.753%	+0.038pp

Allowed numerical tolerance: 0.01 percentage points.

[Copilot]

Copilot review overview

🟢 Approval recommended

The test-only changes deterministically cover the required polling paths and preserve existing runtime behavior.

Review effort: Balanced
Findings: None

What changed in this PR

Stabilizes pet-jsonrpc output polling coverage through deterministic test synchronization.

Changes:

• Adds bounded-channel coordination for pending polling states.
• Tests pending and initially-ready length/error paths.
• Updates existing helper call sites for optional coordination.

File	Description
crates/​pet-jsonrpc/​src/​output.rs	Adds deterministic polling-helper tests without changing production behavior.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[Karthik Nadig (karthiknadig)]

Current-head Copilot review recommends approval with no findings. Native raw LCOV confirms both real polling helpers now execute their pending paths on every platform: length-wait assertion/yield hits are Linux 8, Windows 139, macOS 1; error-wait hits are 9, 491, and 3 respectively.

Coverage against the workflow baselines improves on all three platforms: Linux 85.580% (+0.033pp), Windows 83.130% (+0.055pp), native macOS 80.586% (+0.069pp). All three CodeQL analyses have zero results and no errors/warnings.

I inspected all performance snapshots and exact-inventory checks. Windows/macOS warm discovery improves; Linux is 68ms versus 48ms (+20ms), and cold discovery is 148ms versus 108ms (+40ms). Those differences remain within the unchanged budgets and are accepted as measurement variation for a change wholly inside the test module; no production code or benchmark setup changed. No reruns or relaxed thresholds were used.

The two ordinary macOS test jobs are still finishing, so this remains draft until those gates complete. Required human approval will not be bypassed.