Repository navigation
fix(adaptive): backport same-window idle recovery fix to release/0.8.0 - #15595
Conversation
|
Local test evidence (head Do not merge before #15594. |
Design Review (Fable 5) — ✅ PASSDesign-level review of Design-Verdict: PASS Verify #15594 landed on [DESIGN-REVIEWED] 6c4357e |
First Principles Review (Fable 5) — ✅ PASSPremise-level review of First-Principles-Verdict: PASS Verify the description's own precondition: #15594 merged on What this change shipsInventory (5 items) — 4 justifiedIntent: FIX — on the 0.8.0 line, a due idle-recovery step on the exec cap is no longer delayed by a same-window spawn-gate increase.
[FIRST-PRINCIPLES-REVIEWED] 6c4357e |
GPT 5.6 Review — ✅ no blocking findingsGPT 5.6 completed its review of This comment is updated in place on each push. Review detailsFINDING -- src/kiro_crew/platform_compat.py:7291 -- False positive or not applicable? A repository writer can comment: |
Opus 5 Review — ✅ no blocking findingsReviewed Review detailsThe spec paragraph this PR rewrote still claims every FINDING — docs/system-specs/modules/adaptive-concurrency.md:380 — the unchanged qualifier [OPUS-REVIEWED] 6c4357e Verdict parsed from the review's SHA-scoped output markers for commit False positive or not applicable? A repository writer can comment: |
d4368b3 to
619a513
Compare
|
Re-cut from #15594's amended commit |
619a513 to
54f7708
Compare
|
Opus 5 finding on |
54f7708 to
b0588c0
Compare
|
GPT 5.6 and Opus 5 on |
b0588c0 to
c42ddc9
Compare
|
GPT 5.6 on |
Idle recovery (#15541) took its exec step only when nothing else had changed on the sample. A spawn-gate increase earned on the same sample set that shared flag, so the idle exec step due on it was skipped, and the gate step also restarted the one-increase-per-window clock: the exec cap then waited a further window for a step it had already been due. Judge the idle step on the exec track alone. Both tracks keep sharing one increase-per-window clock across samples, as the earn rules do. An increase's reason now names each track it moved: the exec note only when the exec cap moved, and every gate step as "spawn gate target X -> Y" (the policy's target; the daemon confirms or clamps it). A gate-only step used to read "clean window earned +1", and a gate step beside an exec step was not mentioned at all. (cherry picked from commit f23a39f)
c42ddc9 to
6c4357e
Compare
|
Re-cut onto #15594 |
… release/0.8.0 (#15619) * fix(windows): backport the directory-junction hardening to release/0.8.0 (#15666) Backport of #14327, #14962, #15310 to release/0.8.0, cherry-picked in main's merge order. * fix(windows): anchor six directory-link fences to pinned descriptors (#14327) (cherry picked from commit 6b0a6d5) * fix(config): refuse a link or junction at the workspace dir name (#14962) (cherry picked from commit 4bb50ef) * fix(windows): make four is_symlink guards see a junction (#12379, #10492) (#15310) (cherry picked from commit 6d54165) * style(test): black-format the failed-start reap test and drop its history wording on release/0.8.0 (#15665) test/test_acp_runtime_failed_start_reap.py arrived with #15625 not black-formatted (fails scripts/check_black_formatting.py on the release tip), and the _gone docstring said "no longer runs", which scripts/check_comment_history.py flags as history narration. Black 26.3.1 (--target-version py310) reflow, and the docstring now reads "has stopped running", the wording the #15664 forward-port uses. Test file only. * fix(adaptive): keep a same-window gate step from eating idle recovery (#15595) Idle recovery (#15541) took its exec step only when nothing else had changed on the sample. A spawn-gate increase earned on the same sample set that shared flag, so the idle exec step due on it was skipped, and the gate step also restarted the one-increase-per-window clock: the exec cap then waited a further window for a step it had already been due. Judge the idle step on the exec track alone. Both tracks keep sharing one increase-per-window clock across samples, as the earn rules do. An increase's reason now names each track it moved: the exec note only when the exec cap moved, and every gate step as "spawn gate target X -> Y" (the policy's target; the daemon confirms or clamps it). A gate-only step used to read "clean window earned +1", and a gate step beside an exec step was not mentioned at all. (cherry picked from commit f23a39f) * fix(acp): treat only JSON whitespace as an empty mcp.json (#15673) #15627 read an empty or whitespace-only MCP settings file as absent, using bytes.strip() with no argument. That also strips \x0b and \x0c, which JSON does not allow, so a file holding only those bytes read as "no restrictions" instead of failing closed like any other malformed document. The emptiness check now strips only space, tab, CR and LF (RFC 8259); anything else goes to the parser and fails closed as before #15627. Refs #15616 (cherry picked from commit 157fd51) * fix(sandbox): backport the scope probe PATH and bus-path fixes to release/0.8.0 (#15638) * fix(sandbox): keep the scope probe refresh off PATH, decode bus paths Two defects the post-logout probe fix (#15622) introduced, from its review: - The probe now refreshes on a TTL, and every refresh resolved systemd-run with shutil.which on the event loop. A PATH entry on a stalled mount would hang the gateway's loop. The probe now uses platform_compat.trusted_system_bin, the same fixed-directory lookup cgroup_scope_argv already pins its wrapper with. - DBUS_SESSION_BUS_ADDRESS unix:path= values may percent-escape any byte. Probing the escaped spelling literally misses a live bus that lives outside XDG_RUNTIME_DIR and drops the cgroup ceiling. The value is now decoded before it is dialed. (cherry picked from commit e5a68df) (cherry picked from commit 623f2b4) * fix(sandbox): resolve the probe's systemd-run without any PATH walk Second-round review of this PR: trusted_system_bin's miss diagnostic still calls shutil.which, so a host with no trusted systemd-run walked PATH on the event loop at the first probe refresh. The probe now uses platform_compat.trusted_system_bin_quiet, the same fixed-directory lookup without the diagnostic, and reports the miss in its reason. A D-Bus unix:path= value that decodes to an empty path or one holding a NUL cannot name a socket and made os.stat raise ValueError on the spawn path; such values are now skipped. (cherry picked from commit 5843231) * fix(chat): restore the interrupted turn when Resume lands on a natively loaded session (#15643) (#15644) * fix(chat): restore the interrupted turn when Resume lands on a natively loaded session (#15643) kiro-cli appends a prompt to <sid>.jsonl only once the model has answered it. When the process serving a turn dies mid-answer (a gateway restart, a crash, a recycled runtime) that request is never written, so session/load brings the conversation back without it. A natively resumed session gets no Kiro Crew replay, so the Resume press on that turn sent a bare "finish the user's most recent request" with the request missing: the model answered the turn before it, or said there was nothing to continue. Reproduced against kiro-cli 2.26.1: after a completed turn and a second turn killed mid-stream, session/load restores only the first. The path is any native resume: a dashboard chat with agent.tool_search off, a dashboard chat linked to a channel, and the automatic connection-lost recovery on a respawned runtime. When the Resume message or the connection-lost recovery runs on a session that this turn natively resumed, _run_chat now prepends the interrupted turn from the slot's own transcript (the opener the turn-in-flight marker restores, plus the partial answer), bracketed [INTERRUPTED TURN] so the context scanner attributes it. A cold start that replays already carries the turn and is unchanged. A session whose only turn was interrupted was already covered: SessionMap.get prunes a sid whose log holds no turn. Backport of #15643 to release/0.8.0. Conflict: context.py still carries compress_thread_history here (removed on main by #14952); the new helper is placed after it. * fix(chat): open the interrupted-turn restore at the right row, and mint it after the scrub (#15643) A cron, app or synthesis delivery opens its turn as an inject row. The interrupted-turn walk only stopped at user, nudge and sub-agent rows, so for such a turn it reached the already-answered request before it and restored that as the user's most recent request. Openers now include the inject kinds dashboard.state.TURN_OPENING_INJECT_KINDS names (the dispatching kinds minus recovery and user_replay, which continue a turn). The session-busy recovery (_BUSY_RECOVER_MSG) also resets the session and session/loads the same native conversation, so it gets the restore too. The frame tells the model the request inside it is the one to carry on with, so its markers join _STRUCTURAL_MARKER_RES and the runner prepends it after the egress scrub, like the task checklist; the builder scrubs the restored payload itself. The turn-in-flight marker now copies a sub-agent completion opener too (_LOCAL_TURN_OPENER_ROLES / _LOCAL_TURN_PROMPT_ROLES): without it a crash before the flush loses that row, and the restore would walk past it to the previous, already-answered request. * fix(mcp): repair Windows command casing at every resolver, not only the rewriter On Windows `shutil.which` spells the extension it appends exactly as PATHEXT spells it, so a bare `demo-mcp` resolves to `demo-mcp.EXE` while the file on disk is `demo-mcp.exe`. A launcher that dispatches on its own argv[0] basename case-sensitively (a tool manager's multiplexer shim) refuses to run under that spelling and the MCP server never starts. The rewriter already restored the on-disk basename, but the agent-config resolver (`agent._resolve_command`) and the dashboard probe (`mcp_discovery`) consumed raw `which` output, and the resolver's result is PERSISTED as the spec's absolute `command`. An absolute command is accepted verbatim on every later pass, so the uppercase spelling written once looked operator-authored to the rewriter from then on and its repair never ran. Move the helper to `kiro_crew.env.resolved_command_casing`, beside the `mcp_search_path` the three resolvers already share, and apply it on both return paths of `_resolve_command` and in the probe. Repairing at the resolver also keeps the provenance record's `emitted` value repaired, so `command_is_ours` still recognises the entry and re-derivation stays on. Fixes #11627 (cherry picked from commit aa7c0bd, #15071; applied as its pre-#14919 revision 457210d plus the round-2 test and docstring delta, since release/0.8.0 predates the agent.py split)
Problem / Motivation
Goal: On the 0.8.0 release line, an idle-recovery step on the execution cap is not delayed when the spawn gate earns its own increase in the same window.
Follow-up to #15541, closing the GPT 5.6 finding on its final head
9a56254f8f(the PR merged before the fix could be pushed). InAdaptivePolicy._maybe_increase, the idle exec step was guarded byif not changed.changedis shared with the gate track, so a gate-only increase in the same window made the due idle exec step skip. It also moved_last_increase_at, so the exec cap then waited one more 30 s window for a step it had already earned.Why it matters
While the exec track is idle, MCP backend inits can keep the spawn gate earning. Every gate step would then push the exec recovery back a window, which stretches the "cap returns to the fresh-start value" time #15541 promises, and on a host with steady backend churn it could delay it repeatedly.
Not a goal
What changed (motivation → approach → change)
Symptom: a gate step in the idle window delays the exec idle step. Cause: the idle step read the shared
changedflag rather than the exec track's own. Change: recordexec_changedafter the exec earn rules and guard the idle step on it, so a gate step on the same sample no longer suppresses the idle exec step due on it. Both tracks still share one increase-per-window clock across samples, as the earn rules always have. Every increase's reason now names each track it moved. The exec note appears only when the exec cap moved, and every gate step readsspawn gate target X -> Y on backend inits, the policy's target, since the daemon confirms or clamps it on apply. Before, a gate-only step readclean window earned +1, and a gate step beside an exec step was not named at all. One spec line (Tests) is updated.Backwards compatibility
Compatible: no API, config or state key changes; only the timing of an idle exec step that was already due.
Tests
test_adaptive_policy.py::TestIdleRecovery::test_a_gate_step_does_not_spend_the_idle_exec_step: exec track cut to 1 and idle, gate busy and earning. On the sample where idle recovery is due, both the gate (+1) and the exec cap (1 -> 2) step. The decision's reason names both moves.TestSlowStart::test_an_idle_step_beside_a_doubled_gate_step_reports_both_movespins the slow-start case (gate 4 -> 8, and the idle figure is the clock the rule judged, 60s), andtest_every_gate_step_names_itself_and_only_itselfpins a gate-only step and an earned exec step combined with a gate step. The first test fails onmain(effective_exec_cap == 1, and the gate step is reported asclean window earned +1) and passes with the fix.Manual verification
N/A. This is a pure policy state-machine change, and the unit test drives the exact sample sequence.
Local:
test_adaptive_policy.py,test_adaptive_controller.py,test_resource_status.py,test_overload_acceptance.py,test_adaptive_startup_memory.py,test_mcp_core_more_coverage.pygive 443 passed. flake8, black, mypy (adaptive/) and docs-lint are clean.Related Issues
Backport of #15594. Follow-up to #15541 (GPT 5.6 finding on
9a56254f8f).Pattern harvest
Not generalizable: a single shared "changed" flag across two independently-earning tracks. The file already documents per-track independence in
_reset_bases, and this was the one place that read the combined flag.Checklist
feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...)Contribution License Agreement