You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Lifecycle: the issue exploration reaches new -> held by factory at the act ceiling #202
The base "new, one act short" in tests/test_transitions.py (ISSUE_BASES) cannot reach new -> held by factory at
the act ceiling, a move the Kernel already makes on main. From that base the Controller's one act is accept, which
spends the last act but adds factory:triage, so the issue reads triage and wants a Run, never the act ceiling; every
person's move renews acts_since_writer. To reach the move the issue must spend its last act and still want accept.
It does: a writer's issue in new, one act short, carrying a factory:ready the readiness App added with no readiness
Run that judged it (facts(acts_since_writer=max_acts_per_item - 1, label_events=[labelled(READY, READINESS, 2)])).
The Controller removes that label (remove_label, the last act), and the issue then reads new / act_ceiling / hold: new -> held by factory, (label_not_counted -> act_ceiling), checked by hand in the #171 fix round with a
two-deep exploration of that one base. lifecycle/issue.json does not declare it today, so adding that
base fails the exploration until new -> held by factory, bound max_acts_per_item, is declared (ruling 6 of #171
named it, "if the exploration reaches it"). Proposed follow-up: add that base, declare the move with that bound, and
let the exploration hold it; it fits with the out-of-scope item of #171's ruling 9 that gives the issue Lifecycle's factory Holds a bound.
Done when
The exploration has the base above and reaches new -> held by factory.
lifecycle/issue.json declares it, with bound max_acts_per_item, and the exploration holds it.
Solution Architect, delegated. On main at 9bfc813 (#241). Read from the issue (found by the review of #171), tests/test_transitions.py (ISSUE_BASES, "new, one act short"; controller_does), kernel/issue.py (not_counted: a factory:ready the readiness App added with no Run that judged it is removed; _stop at the act ceiling), lifecycle/issue.json as #232 left it, which declares new → held by the Factory with no bound, and tools/check_lifecycle.py, which asks no bound of a transition into held, since only a person leaves it.
What this story decides. Whether the exploration reaches new → held by the Factory at the act ceiling, and what bound the Lifecycle's row carries. It is decided in #228's ruling 1 and 4, and built in #228's pull request.
1. The base is added to ISSUE_BASES, as the issue states it: a writer's issue in new, one act short, carrying a factory:ready the readiness App added with no readiness Run that judged it. From it the Controller's one act is remove_label, the last act; the issue then reads new / act_ceiling / hold, and the exploration makes new → held by the Factory, (label_not_counted → act_ceiling).
2. The row's bound stays null.lifecycle/issue.json has declared new → held by the Factory since #232, for the Hold the brakes place for unmerged closes (seed 137); the act-ceiling Hold is the same row by another reason. Every → held by the Factory in the issue Lifecycle carries no bound, and tools/check_lifecycle.py needs none: no cycle passes through held without a person, so no bound on the way in is what terminates. A bound of max_acts_per_item would name one of the row's two causes and nothing the check uses; the reason each Hold gives is held by test_a_hold_no_person_makes_names_what_bounds_it, as §4.1's table says. The issue's second box is read as met by the row declared.
3. Record. None beyond #228's row on decision 0028.
4. Files, gates, scope.#228's. The PR says "Closes #202".
Activity
don-petry commentedon Oct 20, 2026
Working rulings (overturnable in review)
Solution Architect, delegated. On main at 9bfc813 (#241). Read from the issue (found by the review of #171),
tests/test_transitions.py(ISSUE_BASES, "new, one act short";controller_does),kernel/issue.py(not_counted: afactory:readythe readiness App added with no Run that judged it is removed;_stopat the act ceiling),lifecycle/issue.jsonas #232 left it, which declaresnew → heldby the Factory with no bound, andtools/check_lifecycle.py, which asks no bound of a transition intoheld, since only a person leaves it.What this story decides. Whether the exploration reaches
new → heldby the Factory at the act ceiling, and what bound the Lifecycle's row carries. It is decided in #228's ruling 1 and 4, and built in #228's pull request.1. The base is added to
ISSUE_BASES, as the issue states it: a writer's issue innew, one act short, carrying afactory:readythe readiness App added with no readiness Run that judged it. From it the Controller's one act isremove_label, the last act; the issue then readsnew / act_ceiling / hold, and the exploration makesnew → heldby the Factory,(label_not_counted → act_ceiling).2. The row's bound stays null.
lifecycle/issue.jsonhas declarednew → heldby the Factory since #232, for the Hold the brakes place for unmerged closes (seed 137); the act-ceiling Hold is the same row by another reason. Every→ heldby the Factory in the issue Lifecycle carries no bound, andtools/check_lifecycle.pyneeds none: no cycle passes throughheldwithout a person, so no bound on the way in is what terminates. A bound ofmax_acts_per_itemwould name one of the row's two causes and nothing the check uses; the reason each Hold gives is held bytest_a_hold_no_person_makes_names_what_bounds_it, as §4.1's table says. The issue's second box is read as met by the row declared.3. Record. None beyond #228's row on decision 0028.
4. Files, gates, scope. #228's. The PR says "Closes #202".
Needs the Maintainer? None.
Exploration: the issue's moves the world makes, and the world held to…