fix(sa-implement): keep daemon alive on empty backlog in continuous mode - #15
Merged
Merged
Conversation
Merge develop into main
The long-running daemon is meant to idle-poll an empty backlog (~15m) and terminate only on an explicit user stop, the kill-switch, a time/PR/escalation budget, or single-shot --once. In practice it exited the moment the backlog drained in normal continuous mode. daemon_loop calls auto-gate.sh with --no-backoff, which makes the gate emit "STOP backlog-empty" on an empty queue. The daemon treated any gate STOP as terminal and returned, so it exited as soon as the backlog was empty; its own idle-poll branch was correct but unreachable. Fix it daemon-side (the daemon owns cadence): in the gate-stop handler, treat "backlog-empty" as NON-terminal when --once was not passed, falling through to the existing idle-poll. Every other stop reason (kill-switch/time/budget/ operator) stays terminal, and --once still stops on an empty backlog. A gate-side flag was rejected by review as incomplete (a second STOP backlog-empty fires after the gate's idle-backoff), so the gate's contract is left unchanged. Also clarify the backlog-empty wording in SKILL.md (terminal only under --once) and correct the stale gh_queue_list doc comment (it accepts an assignee arg). Verified with a deterministic harness (no live GitHub): the real auto-gate.sh decision table (8/8) and the real daemon_loop driven through its FIFO protocol (fixed-mode 10/10, incl. a full idle -> work -> idle -> explicit-stop lifecycle); the pre-fix code reproduces the exit-on-empty bug. Closes #14
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Fixes the
sa-implementdaemon exiting the moment the GitHub issue backlog is empty in normalcontinuous mode. It now idle-polls an empty backlog and stays alive until an explicit user stop,
the kill-switch, a time/PR/escalation budget, or single-shot
--once.What this PR does
bin/auto-daemon.sh— indaemon_loop's gate-stop handler, treatbacklog-emptyasnon-terminal when
--oncewas not passed, falling through to the existing idle-poll. Everyother stop reason stays terminal;
--oncestill stops on empty. The gate's contract isunchanged (a gate-side flag was rejected by review as incomplete — a second
STOP backlog-emptyfires after the gate's idle-backoff).SKILL.md— clarifybacklog-emptyis terminal only under--once.lib/gh.sh— correct the stalegh_queue_listdoc comment (it accepts an assignee arg).Verification
Deterministic harness (no live GitHub): real
auto-gate.shdecision table 8/8; realdaemon_loopdriven through its FIFO protocol 10/10 in fixed-mode (incl. a fullidle → work → idle → explicit-stop lifecycle). Pre-fix code reproduces the exit-on-empty bug.
Closes #14