Repository navigation
CI and test hardening: no timing assertions, no network in unit tests, release gate trusts CI and caches builds - #48
Merged
Conversation
… no network in unit tests, release gate trusts CI and caches builds - bootstrap canary: the default suite checks behaviour only; the timing claim is an #[ignore]d test run by hand - git timeout test hangs on a local ssh listener instead of github.com - integration readiness/drain bounds sized for a shared 2-vCPU runner - release.yml: the gate waits for and requires the CI run on the tagged commit instead of re-running the suite cold (13 min); per-target rust-cache for the release profile so DuckDB is not rebuilt every tag Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KLsmJHt8RwDUfoN7iaemBy
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Why
Two flaky tests failed CI three times today, and a release takes ~45 minutes: the release workflow re-runs the full suite cold (13 min) and then compiles bundled DuckDB from scratch on three runners (13–15 min each) with caching deliberately off. PR CI itself is ~3 minutes.
Tests
engine::tests): the default suite now checks the behaviour (LIMIT 0 creates the empty table with the SELECT's schema). The timing claim — every shape of it failed on a loaded 2-vCPU runner while the gate held — is an#[ignore]d test run by hand when the planner or DuckDB moves.run_gittimeout test hung ongithub.com; it now hangs on a local ssh listener that never answers. No network in unit tests.servereadiness (20 s → 60 s) and SIGTERM drain (5 s → 30 s) bounds sized for a shared runner; the log sentences, not the clock, prove the drain ran.Release workflow
make release) and requires success. Same safety property, 13 minutes less.Swatinem/rust-cachekeyed on the release profile and target, so the dependency tree (DuckDB above all) persists across tags. A warm release linksdatamk; a dependency bump rebuilds.Expected release time after this lands and the caches are seeded once: roughly 5 minutes plus the macOS runner's link.
Tests: 914 unit (1 ignored) + 33 integration, clippy clean.
🤖 Generated with Claude Code
https://claude.ai/code/session_01KLsmJHt8RwDUfoN7iaemBy