Add Litestream checkpoint configuration settings

[epompeii]

Added 3 optional fields to JsonLitestream: min_checkpoint_page_count, checkpoint_interval, and truncate_page_n.

Default truncate_page_n to 0.

[github-actions]

🤖 Claude Code Review

PR: #727
Base: devel
Head: claude/litestream-checkpoint-config-Ga0DL
Commit: c788feab106b683a3de293338e79a328ed2ebc06

---

PR Review: Litestream Checkpoint Configuration

Summary

3 commits adding Litestream checkpoint configuration options (JsonCheckpoint struct) with interval, min_page_count, and truncate_page_n fields, plus documentation updates across 9 languages and OpenAPI spec changes.

---

Issues

1. Breaking YAML output change (Medium)
LitestreamDb::truncate_page_n is non-optional (u64, not Option<u64>), so truncate-page-n: 0 is now always emitted in every generated Litestream YAML — even when the user didn't configure any checkpoint settings. This is a behavioral change for all existing deployments. While 0 is the Litestream default (TRUNCATE disabled), explicitly injecting a new key into the config file could be surprising. Consider either:

• Making truncate_page_n an Option<u64> with skip_serializing_if (only emit when explicitly set), or
• Documenting this as an intentional change that pins the safe default.

2. Missing trailing newlines in docs files (Low)
All 9 localized .mdx files end with \ No newline at end of file. This can cause diff noise in future edits and some tools warn about it.

3. No validation of duration strings (Low, pre-existing pattern)
interval is Option<String> with no validation — a typo like "5x" would only fail at Litestream runtime. This matches existing fields (sync_interval, snapshot_interval), so it's consistent, but worth noting as technical debt.

---

Positive Observations

• Good use of destructuring in into_yaml — matches project convention and ensures compiler errors if fields are added later.
• TRUNCATE_DISABLED constant is well-documented and provides a clear semantic name for the magic value 0.
• Thorough test coverage — new into_yaml_with_checkpoint_config test covers the happy path, and existing tests are updated to assert the new default behavior.
• OpenAPI spec updated via cargo gen-types — stays in sync with the Rust types.
• Consistent patterns — JsonCheckpoint follows the same Option + skip_serializing_if pattern used throughout the config module.
• Documentation updated across all 9 languages simultaneously.

---

Verdict

The change is clean and well-structured. The main point to address is issue #1 — whether truncate-page-n: 0 should always appear in generated YAML or only when explicitly configured. The missing trailing newlines (#2) are a minor fix.

---

_{Model: claude-opus-4-6}

[github-actions]

Bencher Report

Branch	claude/litestream-checkpoint-config-Ga0DL
Testbed	ubuntu-22.04

Click to view all benchmark results

Benchmark	Latency	Benchmark Result microseconds (µs) (Result Δ%)	Upper Boundary microseconds (µs) (Limit %)
Adapter::Json	📈 view plot 🚷 view threshold	3.78 µs (+6.36%) Baseline: 3.56 µs	4.69 µs (80.70%)
Adapter::Magic (JSON)	📈 view plot 🚷 view threshold	3.74 µs (+5.98%) Baseline: 3.53 µs	4.61 µs (81.10%)
Adapter::Magic (Rust)	📈 view plot 🚷 view threshold	25.00 µs (-1.99%) Baseline: 25.50 µs	31.53 µs (79.28%)
Adapter::Rust	📈 view plot 🚷 view threshold	2.82 µs (+0.26%) Baseline: 2.81 µs	3.25 µs (86.70%)
Adapter::RustBench	📈 view plot 🚷 view threshold	2.80 µs (-0.31%) Baseline: 2.81 µs	3.25 µs (86.22%)
head_version_insert/batch/10	📈 view plot 🚷 view threshold	102.62 µs (+1.91%) Baseline: 100.70 µs	118.47 µs (86.62%)
head_version_insert/batch/100	📈 view plot 🚷 view threshold	242.67 µs (+1.85%) Baseline: 238.26 µs	265.15 µs (91.52%)
head_version_insert/batch/255	📈 view plot 🚷 view threshold	465.87 µs (+0.57%) Baseline: 463.25 µs	498.73 µs (93.41%)
head_version_insert/batch/50	📈 view plot 🚷 view threshold	162.34 µs (+0.56%) Baseline: 161.44 µs	182.69 µs (88.86%)
threshold_query/join/10	📈 view plot 🚷 view threshold	151.39 µs (+4.17%) Baseline: 145.33 µs	169.23 µs (89.46%)
threshold_query/join/20	📈 view plot 🚷 view threshold	162.99 µs (+1.95%) Baseline: 159.87 µs	186.12 µs (87.57%)
threshold_query/join/5	📈 view plot 🚷 view threshold	147.52 µs (+7.32%) Baseline: 137.46 µs	160.02 µs (92.19%)
threshold_query/join/50	📈 view plot 🚷 view threshold	199.44 µs (-0.90%) Baseline: 201.26 µs	231.06 µs (86.31%)

🐰 View full continuous benchmarking report in Bencher