A loaded playbook now decides a real task's workflow. The wiring point is the daemon's definition pin (Daemon.resolveWorkflowID, consumed by pinWorkflowFromCollection), not pollNATS/publishTask as the design doc assumed: routing moved out of the agent process when workflow definitions became database rows pinned before dispatch, so the pin is the one place a task's workflow is chosen. Precedence: an explicit Task.Workflow (the waiting_human -> approved requeue) still wins, then a matching playbook, then workflow.ResolveWorkflowID's existing label/kind bindings. A playbook naming an undefined workflow is a reported failure that parks the task rather than a silent fall-through to the binding's choice -- the operator bound that trigger deliberately. Store.Dispatch returns the selected workflow name plus the matching playbook's id and version, instead of a compiled workflow.Workflow: the production caller decides against the active definition collection, and the id/version are the first two components of the resolved gap-2 idempotency key. The playbook package no longer imports internal/domain/workflow. The shipped example's `when` read event.priority, a field no intake supplies, so it could never have fired once wired. Both examples now read the event surface playbookInput really builds, and label-trigger.yaml documents the axis an instance defines for itself: trigger.kind is the closed workintake vocabulary because a kind addresses a NATS subject, while trigger.labels is free-form and binds an operator's own taxonomy to a workflow. Module/Channel/Forge positions stay rejected at load; they need the playbook_dispatches ledger first (archie-core-t2db.24). Closes archie-core-t2db.23
50 lines
2.2 KiB
YAML
50 lines
2.2 KiB
YAML
# Example EDA playbook document (t2db.15 slice).
|
|
#
|
|
# Install: copy this file into the configured eda_playbook_dir (default
|
|
# ~/.config/archie/eda-playbooks) and restart archied. It is loaded at
|
|
# startup with the reject-at-load rule: any malformed playbook, multi-action
|
|
# playbook, non-workflow action position, or `when` compile failure aborts
|
|
# startup with a reported error -- a partially-valid set never runs.
|
|
#
|
|
# This slice supports exactly ONE action per playbook, position: workflow.
|
|
# Multi-action playbooks and Module/Channel/Forge action positions stay
|
|
# blocked on the unresolved execution-time gaps and are rejected at load.
|
|
#
|
|
# A trigger matches on either axis, and both are checked when both are given:
|
|
#
|
|
# kind: the closed workintake routing vocabulary -- bug, feature,
|
|
# bootstrap, default. It is closed because a kind also addresses the
|
|
# task's NATS subject; anything else is rejected at load. Set it
|
|
# when you want the rule to follow archie's own classification.
|
|
# labels: any labels you like, including ones no kind recognises
|
|
# ("ops::security-review"). All of them must be present on the
|
|
# issue. This is the axis an instance defines for itself: it binds
|
|
# your own taxonomy to a workflow with no code change. See
|
|
# label-trigger.yaml. The `when` condition is a CEL
|
|
# expression over `event` -- see docs/prds/eda-playbook-engine.md, resolved
|
|
# open question 1.
|
|
#
|
|
# A matching playbook decides the task's workflow ahead of the kind/label
|
|
# binding files, and a workflow it names that is not defined parks the task
|
|
# rather than falling back. For a forge-discovered task, `event` carries:
|
|
#
|
|
# kind routing kind the labels produced (bug/feature/bootstrap/default)
|
|
# labels the issue's labels, as a list
|
|
# owner repo owner
|
|
# repo repo name
|
|
# number issue number
|
|
# title issue title
|
|
# body issue body
|
|
# source "forge" or "chat"
|
|
#
|
|
# Reading a field that is not in that set is an evaluation error, which the
|
|
# coordinator treats as the condition being false (the playbook is skipped).
|
|
|
|
trigger:
|
|
kind: bug
|
|
|
|
actions:
|
|
- position: workflow
|
|
workflow: tdd
|
|
when: '"regression" in event.labels'
|