Every ticket as the root of its provenance graph — goal, patch, approval, runs and (as the ticket-class build lands) mandate, manifestations, guarantees, journey, artifacts and children. Rendered through the <<<TicketGraphList>>> tag.
STOP: #265 — forge rewrite unit names a source its worker cannot read (#346)
goal
HALT #265. It cannot be executed as filed, and its body cannot be repaired.
MODULE : gitlab__testmonkeyalpha-group__jazz-forge__audio_tagger:tests/test_audio_tagger.py
MODULE KEY : forge-rw:24ded84a0ff2941e
SUPERSEDED BY : #353 — the same module, the same requirements, plus that module's ORIGINAL
SOURCE embedded verbatim in the body. THE WORK HAPPENS THERE.
WHY THIS UNIT IS HALTED. #265 names its module's source only as a
`git --git-dir=<forge>/remotes/<repo>.git cat-file blob <sha>` command. The pair_loop WORKER
that executes these units is spawned into the ticket's own congruency worktree and is
SANDBOXED to it: the forge's bare repos are outside that directory, so it cannot run that
command, and no cached copy exists. The unit blocks with "canonical source unreadable,
refused to fabricate" — which is the worker behaving CORRECTLY, because #200 demands
behavior PRESERVED and a source you cannot read cannot be preserved. The pilot #259 proved
it end to end. Ticket #346 is the fix, and #353 above is this module's share of it.
WHY A STOP AND NOT AN EDIT. A ticket's body is PERMANENT in SQL: tg_tickets_immutable_
identity ABORTs any UPDATE that changes id/ts/title/body/component/kind
(migrations/0029_ticket_authorization.sql:111-117), and no route edits a body. So #265's body
cannot be given the source it lacks. The schema's own remedy for exactly this case is a
GRAFT — "stop + spawn + supersedes-edge" (0029_ticket_authorization.sql:8) — and that is
what happened: #353 was filed carrying the source, a `supersedes` edge runs from #353 to #265,
and this ticket is the stop.
NOTHING WAS DELETED. #265, its journal and its edges stay on the record forever; it is halted,
not erased. To release it, aim a `start` ticket at THIS ticket through a `starts` edge (a
start is aimed at the stop ticket it releases — migrations/0030_pair_loop_routes.sql).
Releasing it hands a sandboxed worker the same unreadable unit again, so prefer #353.
Filed mechanically by checkouts/current/congruency/tools/forge_refactor_ticketer.py under ticket #346.
patch
none
approval
unapproved
runs
no runs recorded
mandate (clauses)
not yet recorded — lands with the ticket-class build
manifestations
not yet recorded — lands with the ticket-class build
guarantees
not yet recorded — lands with the ticket-class build
journey (blunders & successes)
not yet recorded — lands with the ticket-class build
artifacts (forge)
not yet recorded — lands with the ticket-class build