docs(eda): record the shipped action-playbook rules and stale claims

This commit is contained in:
sam committed 2026-09-22 09:51:40 +10:00
1 parent 4c556113eb
commit 38ddfbb001
4 files changed
+18 -13

No files matched your search

+1 -1
View File
@@ -21,7 +21,7 @@
# workflow_routing_file = "~/.local/share/archie/workflow-routing.yaml" # kind->workflow rebinding (bug/feature/bootstrap)
# workflow_labels_file = "~/.local/share/archie/workflow-labels.yaml" # arbitrary forge label -> workflow (EDA playbook slice 2)
# playbook_dirs = ["~/.local/share/archie/workflows"] # list of dirs of *.yaml binding files, merged with the two above (slice 3); a key bound twice across sources is a load failure
# eda_playbook_dir = "~/.local/share/archie/eda-playbooks" # rich EDA playbook documents (trigger + CEL gated single workflow action); a matching one decides the task's workflow ahead of the three binding sources above; distinct from playbook_dirs
# eda_playbook_dir = "~/.local/share/archie/eda-playbooks" # rich EDA playbook documents (one trigger; either a CEL-gated workflow action or a list of module actions); a matching workflow playbook decides the task's workflow ahead of the three binding sources above; distinct from playbook_dirs
poll_interval = "60s"
label = "archie"
+8 -5
View File
@@ -66,11 +66,15 @@ naming the playbook and the offending action, for:
must be a lowercase CEL field name writable as `actions.<id>` (so a CEL
keyword such as `in`, `true`, `false`, or `null` is rejected at load);
- an `actions.<id>` read of an id no earlier action declares;
- an `actions` read that is not statically resolvable to an id (a dynamic
index, `in`, a comprehension — already enforced);
- an `actions` read that is not statically resolvable to an id (any map index,
literal or dynamic, `in`, a comprehension): the typed `actions` root is an
object type, so only field selection `actions.<id>` resolves and the
`actions["id"]` spelling is not accepted;
- a `result` field the referenced kind's `Result` does not define;
- a reference to an earlier action that produces no result;
- an unknown `position`, an unknown `kind`, or a `module` action with no `kind`;
- a field belonging to the other shape (`kind` on a `workflow` action, or
`workflow` on a `module` action);
- an arg KEY the kind's `Args` schema does not define. Arg VALUE type-checking
is tracked as `archie-core-t2db` debt, not shipped here.
@@ -83,9 +87,8 @@ type-checking is tracked as `archie-core-t2db` debt.
Result typing depends on the ids a playbook declares and the kind each id runs,
so the CEL environment is built from one playbook's action list rather than once
per `Store`. `expr.NewEnv()` declares `actions` as `map(string, dyn)` and takes no
parameters, so its signature changes to accept the declared ids and their result
types.
per `Store`, and `expr.NewEnv` takes that list: each declared id becomes a field
of the `actions` object typed by its kind's `Result`.
## Running an action playbook is out of scope
+5 -4
View File
@@ -29,16 +29,17 @@ linear evaluation when macros are bounded, cost-limited).
```yaml
actions:
- position: module
- id: build
position: module
kind: log
args:
message: '"build finished"' # literal
when: 'event.label == "bugfix"' # gate on event field
- position: module
kind: notify
kind: log
args:
message: '"priority " + string(event.priority)'
when: 'actions.notify.result.delivered == true'
message: '"done: " + string(actions.build.result.written)'
when: 'actions.build.result.written == true'
```
## Trust boundary of the expression
+4 -3
View File
@@ -8,9 +8,10 @@
// every program. `actions` is not a global map: its shape depends on the ids
// one playbook declares and the kind each id runs, so NewEnv takes the
// declared ids and their Go Result struct types and builds a per-playbook
// object type (multi-action-playbooks.md, D3). The same compile path serves
// the playbook loader (reject-at-load) and the lint tool, so author-time
// diagnostics and runtime evaluation cannot disagree.
// object type (multi-action-playbooks.md, D3). The playbook loader compiles
// every `when` and `args` value here, so load-time rejection and runtime
// evaluation cannot disagree; the standalone lint command covers the flat
// kind/label binding files, not rich EDA playbook documents.
//
// CEL is non-Turing-complete, side-effect-free, and panic-free by design (no
// recover() wrapper is needed; verified in the t2db.14 acceptance tests