Problem
crates/mds-cli/tests/cli_watch.rs:1152 — watch_ctrl_c_prints_stopped_watching is timing-flaky in CI. It reded out the v0.3.0 release PR CI (Rust — fmt, clippy, test: 56 passed, 1 failed) on a commit that only touched ci.yml (no Rust change), and passed on re-run. Part of the known-flaky watch-test family.
thread 'watch_ctrl_c_prints_stopped_watching' panicked at crates/mds-cli/tests/cli_watch.rs:1152:5
test result: FAILED. 56 passed; 1 failed
Impact
Intermittent red CI on unrelated PRs; forces manual re-runs (it blocked the v0.3.0 release until re-run).
Fix
Make the Ctrl-C / "stopped watching" assertion deterministic — poll-with-timeout on the expected output / synchronize on the process signal handling rather than relying on fixed timing — or quarantine if it can't be made reliable.
Found during the v0.3.0 release.
Problem
crates/mds-cli/tests/cli_watch.rs:1152—watch_ctrl_c_prints_stopped_watchingis timing-flaky in CI. It reded out the v0.3.0 release PR CI (Rust — fmt, clippy, test: 56 passed, 1 failed) on a commit that only touchedci.yml(no Rust change), and passed on re-run. Part of the known-flaky watch-test family.Impact
Intermittent red CI on unrelated PRs; forces manual re-runs (it blocked the v0.3.0 release until re-run).
Fix
Make the Ctrl-C / "stopped watching" assertion deterministic — poll-with-timeout on the expected output / synchronize on the process signal handling rather than relying on fixed timing — or quarantine if it can't be made reliable.
Found during the v0.3.0 release.