mirror of
https://github.com/kuhyx/testsAndMisc.git
synced 2026-11-12 08:31:03 +01:00
Krzysztof kuhy Rudnicki
8ab15446fd
`git push` took ~130s because ci-mirror re-ran `pre-commit --all-files` plus every lib/tests suite on every push, regardless of what changed. The larger problem found while measuring it: ci_mirror.sh invoked pytest_changed_packages.py with no arguments, and that script returns 0 on empty argv because it is normally a pre-commit hook fed staged paths. The pre-push python stage has therefore been running ZERO tests and reporting success. Behind it sat a genuinely failing test, fixed inff3ea99a. The stage is now passed the changed paths and actually runs. Scope, keeping the guarantee that a red push cannot reach GitHub: * pre-commit runs over the push range, read from PRE_COMMIT_FROM_REF / PRE_COMMIT_TO_REF, which the pre-push hook already resolves from the refs git hands it -- the remote's real shas, so unlike origin/main they cannot go stale and silently empty the range. * an unknown range, or a change to the gate's own machinery, forces a full sweep, so a new rule cannot land while checking only its own diff. * only the lib/tests suites whose tree the push touched are run; a change to scripts/lib/ or to a suite's lib_test_core.sh fans out to all of them. * pre-commit and pytest run concurrently. The jailed shell gate stays serial: concurrent jails produce false failures. * an unchanged tree short-circuits on a recorded tree hash. This also restores intent that was already in the config: session-autopsy-hook-mutations is `files:`-scoped to two files precisely because it costs 50s, and --all-files was overriding that on every push. Measured, against a ~130s baseline: docs-only push 7.6s python push 25.3s (includes a 22s stage that never ran) one shell suite 43.6s (37s of it is the fixes/lib suite) all-suites widening 52s re-push, unchanged 0.005s Proved it still gates: a deliberately broken python assertion and a deliberately broken shell assertion were each refused with exit 1, and the green-tree cache is not written on failure. Both probes reverted. Split into ci_mirror_range.sh and ci_mirror_shell.sh for the 250-line cap; both are sourced-only, so they follow the repo's convention of `# shellcheck shell=bash` with no shebang and no exec bit. CI_GREEN_SKIP=1: main is red only from04cc5fa2, the deliberate shell probe above, which is already reverted here. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01PJrqKERAo5QUeEHWS5Asgp
48 lines
3.3 KiB
JSON
48 lines
3.3 KiB
JSON
{
|
|
"intent": "Cut the pre-push gate from ~130s to seconds by checking the push range instead of the whole tree, and make the pytest stage actually run -- it had been silently executing zero tests on every push.",
|
|
"scope": [
|
|
"meta/scripts/ci_mirror.sh",
|
|
"meta/scripts/ci_mirror_range.sh (new)",
|
|
"meta/scripts/ci_mirror_shell.sh (new, split for the 250-line cap)",
|
|
"Non-goal: weakening any check. The guarantee that a red push cannot reach GitHub is preserved."
|
|
],
|
|
"changes": [
|
|
"FIX: ci_mirror.sh invoked pytest_changed_packages.py with no arguments; that script returns 0 on empty argv, so the pre-push python stage ran zero tests and reported success. It is now passed the changed paths and genuinely runs.",
|
|
"Scope pre-commit to the push range via PRE_COMMIT_FROM_REF/PRE_COMMIT_TO_REF, which the pre-push hook already resolves from the refs git supplies. Repo-wide hooks (file-length-cap --all, no-binaries, check-python-location) are always_run in the config and still sweep the tree.",
|
|
"Force a full sweep when the range is unknown or when the gate's own machinery changed (.pre-commit-config.yaml, meta/scripts/, requirements.txt, .githooks/).",
|
|
"Run only the lib/tests suites whose tree the push touched; a change under scripts/lib/ or a suite's lib_test_core.sh/lib_test_path.sh fans out to all suites.",
|
|
"Run pre-commit and pytest concurrently; keep the shell/jail gate serial to avoid the CPU-contention false failures documented in the handoff.",
|
|
"Short-circuit on a tree hash matching the last green run, recorded in .ci-mirror-venv/ (already gitignored, so invisible to git status and jail fingerprints)."
|
|
],
|
|
"verification": [
|
|
{
|
|
"command": "PRE_COMMIT_FROM_REF/TO_REF set; time bash meta/scripts/ci_mirror.sh",
|
|
"result": "pass",
|
|
"evidence": "docs-only push 7.6s; python push 25.3s (now including a 22s pytest stage that never previously ran); single shell suite 43.6s; all-suites widening 52s; unchanged-tree re-push 0.005s. Baseline was ~130s."
|
|
},
|
|
{
|
|
"command": "Deliberately broke a python assertion, committed, ran the gate",
|
|
"result": "pass",
|
|
"evidence": "Gate exit 1, reported the failing test, and did NOT write the green-tree cache. Reverted."
|
|
},
|
|
{
|
|
"command": "Deliberately broke _t_eq in utils/lib/tests/test_study_doc_lookup.sh at a pinned range",
|
|
"result": "pass",
|
|
"evidence": "Suite failed bare (exit 1) first; gate then exit 1 with 'lib/tests suite failed: .../utils/lib/tests/run_all.sh'. Reverted."
|
|
},
|
|
{
|
|
"command": "shellcheck -x and shfmt -l on all three scripts; check_file_length.sh --all",
|
|
"result": "pass",
|
|
"evidence": "shellcheck clean, shfmt clean, all files within the 250-line cap (232/78/119)."
|
|
}
|
|
],
|
|
"risks": [
|
|
"Scoping means a hook no longer re-checks files outside the push range. Mitigated by the gate-widening rule and by keeping repo-wide hooks always_run.",
|
|
"The green-tree cache could mask a gate change; mitigated because the cache key is the whole-repo tree hash, so any gate edit invalidates it."
|
|
],
|
|
"rollback": [
|
|
"git revert the commit; ci_mirror.sh returns to --all-files on every push (~130s) and the pytest stage returns to silently running nothing.",
|
|
"Delete .ci-mirror-venv/.last-green-tree if a stale cache is ever suspected."
|
|
]
|
|
}
|