Design 47 G: count both run bounds from the ask's taking, 30 minutes

A bound counted from the asking would end a run that only waited in the queue as could not
run, which is not true; the queue is the build-capacity finding's to time (A-466n).
This commit is contained in:
2026-11-17 10:33:58 +01:00
parent 661fb8a742
commit cebac2e791
2 changed files with 9 additions and 5 deletions
@@ -1575,9 +1575,10 @@ the principle that settles it, cited as "design 47 G A-n".
moment it started, so a later step of the same candidate reuses that verdict knowingly, within the candidate.
- **A-466m** (§729; principle 1) The runs' asks are named in `mesh-wants`' `step-progress`, and the watchman times a
composition step from there: the runner says `step-started` before any run is chosen, so the part that knows says it.
- **A-466n** (§720, §725; principle 11) A run's bound is a setting: the assigner's `composition-step-bound`, default 30
minutes (the step bound of 2026-10-26), and the runner's `test-step-bound` per level, its default that level's
existing run bound.
- **A-466n** (§720, §725; principles 11 and 3) A run's bound is a setting, the assigner's `composition-step-bound` and
the runner's `test-step-bound`, each default 30 minutes counted from the build seat's `ask-taken` of the run's ask; an
ask still in the queue is timed by §86's build-capacity finding, never by this bound, since a false `could not run`
would say what is not true.
- **A-466o** (§730, §732; principle 3) After a return to an earlier build, the watchman's `pipeline-stands` says a
reader left standing; the watchman on main timed only a build seat's running ask, so §732 makes it time every step
that reads a run.
@@ -69,7 +69,9 @@ domain of 2026-11-17 was answered by the lead in A-466 to A-466l (G-22), each ci
bound key with its time to live, A-77, A-101; A-466n) ends `could not run`, naming the ask and the bound, and every reader
with it. The end is heard on the assigner's durable consumer of the build seat's `step-ended` (A-240); the assigner
never reads the build seat's state (ADR 0596). A composition step's bound is the assigner's setting
`composition-step-bound`, default 30 minutes (the step bound of 2026-10-26).
`composition-step-bound`, default 30 minutes, counted from the build seat's `ask-taken` of the run's ask, never
from its asking: an ask still waiting in the queue is timed by the build-capacity finding of §86, not by this
bound, since a `could not run` for a run that only waited for a build node would say what is not true (A-466n).
- §721 **A machine's verdict names the modules it blames.** (ADR 0666 §17; A-452, A-466g, A-466h; principles 10, 3, 7
and 11) The composer's verdict of a machine (mesh-delivery-sdk's `composer.Machine`, answered by the node-engine's
`mesh-compose`) gains `blamed`, a list of `{module, problem}`.
@@ -143,7 +145,8 @@ domain of 2026-11-17 was answered by the lead in A-466 to A-466l (G-22), each ci
The runner hears the run's end as the build seat's `step-ended` of its `ask`, on its durable consumer, as it hears
every step (§24). A run whose `ask` has no `step-ended` within the bound of the step that asked it ends `could not
run`, as §720 has it (A-466n), and every reader with it. A test step's bound is the runner's setting
`test-step-bound`, one per level, its default its level's existing run bound. At its start the runner takes up again each key with no `ended`, by its
`test-step-bound`, default 30 minutes, counted from the build seat's `ask-taken` of the run's ask as §720 counts the
composition step's; an ask still in the queue is timed by §86's build-capacity finding alone (A-466n). At its start the runner takes up again each key with no `ended`, by its
`ask` and that bound. It deletes a key once the time its setting `test-runs-kept`, default 7 days, names has passed
since its `ended`, in the step log's order as §330 ages its `pipelines`, and deletes the run's output object with the
key, after the last reader's end. At the cap no run is asked, and the step ends `could not run`, naming the fill, as