Found while landing #144: tests_v2/** appears in ci.yml only under the pull_request path triggers. No CI job actually executes the python socket suites (tests/ or tests_v2/); they only run via the dispatch-only e2e workflow or manually on the VM.
Consequence: regression tests like tests_v2/test_workspace_split_title_updates.py (#144) never gate a PR. The two-commit red/green policy is invisible in CI for socket-level fixes.
Scope: add a CI job that boots a tagged debug build on the runner (or reuses the e2e harness) and runs ./scripts/run-tests-v2.sh, or a curated stable subset if the full suite is too slow/flaky for PR gating. Needs the usual runner budget care (see compat-tests timeout history).
Found while landing #144:
tests_v2/**appears in ci.yml only under the pull_request path triggers. No CI job actually executes the python socket suites (tests/ or tests_v2/); they only run via the dispatch-only e2e workflow or manually on the VM.Consequence: regression tests like tests_v2/test_workspace_split_title_updates.py (#144) never gate a PR. The two-commit red/green policy is invisible in CI for socket-level fixes.
Scope: add a CI job that boots a tagged debug build on the runner (or reuses the e2e harness) and runs ./scripts/run-tests-v2.sh, or a curated stable subset if the full suite is too slow/flaky for PR gating. Needs the usual runner budget care (see compat-tests timeout history).