# v0.42.0 — Documentation Truthfulness, Test Quality & Fuzz Automation > **Full technical details:** [v0.42.0.md-full.md](v0.42.0.md-full.md) **Status: Planned** | **Scope: Large** > Implement the missing `repair_stream_table` function, rewrite the catalog > generator, fix all stale docs, replace 116 fixed sleeps with state-polling > helpers, add differential aggregate property tests, and automate fuzz targets > in CI. --- ## What is this? Two complementary assessment dimensions — documentation drift and test quality — are addressed together. Both share the same root cause: code moves faster than the surrounding infrastructure (docs, tests, CI) that keeps it honest. Fixing one without the other leaves the project looking more trustworthy than it is. --- ## Implement `repair_stream_table` and fix catalog generation Backup/restore docs tell users to call `pgtrickle.repair_stream_table(name)` after restore — but the function does not exist. v0.42.0 implements it and gates the restore runbook with an E2E test. Alongside it, the regex-based catalog generator is replaced with Rust-aware extraction, both catalogs are regenerated, and a CI check prevents future drift. --- ## SQL reference freshness and stale-term audit `create_stream_table` docs omit recent parameters, deprecated GUCs appear as active tuning knobs, and architecture docs reference old single-file modules. v0.42.0 fixes all of these, adds `just docs-lint`, and gates generated reference freshness in CI. --- ## Replace fixed sleeps with state-polling helpers 116 fixed sleeps across the E2E suite make tests slow and flaky. v0.42.0 introduces `wait_for_refresh_history`, `wait_for_cdc_mode`, `wait_for_scheduler_tick`, and `wait_for_job_status` helpers and replaces the highest-impact sleeps across WAL CDC, scheduler, cascade, and bgworker tests. --- ## Aggregate correctness property tests and fuzz CI `SUM(CASE ...)` has only a FULL-mode coverage example; the GROUP_RESCAN fallback can regress silently. v0.42.0 adds differential E2E tests and upgrades CASE detection to the parsed AST level. FULL JOIN aggregate property tests compare DIFF vs FULL across multi-cycle sequences. Fuzz targets finally run in CI with a scheduled smoke job and corpus replay on PRs. --- ## Also in v0.42.0 - WAL decoder SQL parameterization for adversarial payload safety - Stale EC-06 comments replaced with current keyless design notes - Keyless multiset property test --- ## Scope v0.42.0 is a large release, merging what were formerly two separate versions. The work is coherent: both docs and test quality are trust infrastructure that should land together before moving to performance and security hardening. > **Source:** [plans/PLAN_OVERALL_ASSESSMENT_9.md](../plans/PLAN_OVERALL_ASSESSMENT_9.md) > findings FEAT-01, DOC-01–08, CQ-03, CQ-04, TEST-01–02, TEST-05–06, TEST-09, > CI-05–07, OPS-05, COR-04, COR-05, COR-07, COR-08. --- *Previous: [v0.41.0 — DVM Correctness: Structural Cache Keys, Placeholder Safety & WAL Transition Guards](v0.41.0.md)* *Next: [v0.43.0 — Performance Tunability, Diagnostics & Refresh Intelligence](v0.43.0.md)*