Skip to content

test: check the lanes' first look by structure, not by a zero-second clock #121

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
addisdev merged 3 commits into from
Oct 27, 2026
Merged

test: check the lanes' first look by structure, not by a zero-second clock #121

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
7572d61
test: check the lanes' first look by structure, not by a zero-second …
claude Oct 26, 2026
91f53b7
test: keep the first-look case out of $(...) for macOS bash 3.2
claude Oct 26, 2026
553d74e
Merge branch 'main' into test/lane-wait-first-look
addisdev Oct 27, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
20 changes: 18 additions & 2 deletions hooks/tests/run.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -962,7 +962,18 @@ start_bg l-1
wait_for "1" "events held" 10
ok "the single lane held the light job" "$(field "$(row l-1 held)" reason)" "1 job(s) already running, at the limit of 1"
ok "...where the lanes would not have" "$(field "$(row l-1 held)" lane_would_hold)" "false"
ok "...from the start of the wait" "$(field "$(row l-1 held)" lane_wait_s)" "0"
# lane_wait_s is WAITED when the lanes first said yes, and the held row is
# written once, at the end of the hook's first pass. So a number there, rather
# than null, is the lanes' answer at the first look. Its value is not checked:
# WAITED starts before the hook resolves its owner and worker log, so a loaded
# runner can cross a second boundary first and record 1 (run 40331626282).
# (The case stays out of $(...): macOS's bash 3.2 cannot parse one in there.)
L1_LANE_WAIT="$(field "$(row l-1 held)" lane_wait_s)"
case "$L1_LANE_WAIT" in
'' | *[!0-9]*) L1_FIRST_LOOK="not recorded ($L1_LANE_WAIT)" ;;
*) L1_FIRST_LOOK=first ;;
esac
ok "...from its first look" "$L1_FIRST_LOOK" "first"
ok "the row says which policy decided" "$(field "$(row l-1 held)" lanes)" "observe"
ok "the single lane's limit is the one reported" "$(field "$(row l-1 held)" limit)" "1"
start_bg h-2
Expand All @@ -975,7 +986,12 @@ ok "the single lane admitted its oldest waiter" \
"$(grep '"event":"admitted"' "$LOG" | tail -1 | sed -n 's/.*"runner":"\([^"]*\)".*/\1/p')" "l-1"
ok "it waited under the single lane" \
"$([ "$(field "$(row l-1 admitted)" waited_s)" -ge 1 ] && echo waited || echo instant)" "waited"
ok "the lanes would have let it in at once" "$(field "$(row l-1 admitted)" lane_wait_s)" "0"
# Still the first look's answer, not one taken when the single lane let it in:
# that would read as the full wait, which is at least a poll past the first.
L1_ADMITTED="$(row l-1 admitted)"
ok "the lanes would have let it in at once" "$(field "$L1_ADMITTED" lane_wait_s)" "$L1_LANE_WAIT"
ok "...before the single lane did" \
"$([ "$(field "$L1_ADMITTED" lane_wait_s)" -lt "$(field "$L1_ADMITTED" waited_s)" ] && echo before || echo "not before")" "before"
end_jobs

echo "== lanes observe with admission observe: would-hold carries the lane view =="
Expand Down
Loading