diff --git a/docs/briefs/r3-t-wad-full-r3-slice-4-shape-agnostic-skeleton.md b/docs/briefs/r3-t-wad-full-r3-slice-4-shape-agnostic-skeleton.md index e1c5c1eddb4..a2956a1856c 100644 --- a/docs/briefs/r3-t-wad-full-r3-slice-4-shape-agnostic-skeleton.md +++ b/docs/briefs/r3-t-wad-full-r3-slice-4-shape-agnostic-skeleton.md @@ -29,8 +29,8 @@ semantic authority. - `dsl/gunbc/ci_emission.dag` once WI-2 lands. - `CIWorkflowDag` authority for gunbc CI topology. -- `EmissionTarget` standalone gunbc sum type. -- `project_github_actions(CIWorkflowDag, EmissionTarget) -> Workflow`. +- `WorkflowRuntime` standalone gunbc sum type. +- `project_github_actions(CIWorkflowDag, WorkflowRuntime) -> Workflow`. - `Expression::OpaqueString` for GitHub Actions expressions that remain opaque until a non-YamlStatic consumer needs structural evaluation. @@ -45,7 +45,7 @@ semantic authority. ## Acceptance -- No `EmissionTarget` field in `dsl/extdeps/github/actions.dag`. +- No `WorkflowRuntime` field in `dsl/extdeps/github/actions.dag`. - No independent hand-authored `Workflow` copy for gunbc CI. - `Expression::OpaqueString` is emitted verbatim to YAML. - Focused T-CI-WAD tests pass. diff --git a/docs/briefs/r3-t-wad-full-r3-emission-target-gunbc-ci-sketch.md b/docs/briefs/r3-t-wad-full-r3-workflow-runtime-gunbc-ci-sketch.md similarity index 67% rename from docs/briefs/r3-t-wad-full-r3-emission-target-gunbc-ci-sketch.md rename to docs/briefs/r3-t-wad-full-r3-workflow-runtime-gunbc-ci-sketch.md index e983edd0dcd..d37d06bcc3c 100644 --- a/docs/briefs/r3-t-wad-full-r3-emission-target-gunbc-ci-sketch.md +++ b/docs/briefs/r3-t-wad-full-r3-workflow-runtime-gunbc-ci-sketch.md @@ -9,10 +9,14 @@ It must not carry gunbc CI emission policy. The CI projection choice lives in the `dsl/gunbc/` namespace and is supplied at projection invocation time. Terminology remains deliberately separate from Shape-A compiler target -selection. `src/v3/SELF_HOSTING.md` says YAML is a Shape-B artifact produced by -`.dag` programs, never a compiler `EmissionTarget` value. The ratified -T-CI-WAD `EmissionTarget` below is a gunbc CI projection mode for workflow -artifact generation, not a compiler target. +selection. `src/v3/SELF_HOSTING.md:609` defines the Shape-A +`type EmissionTarget { language: LanguageSpec, rendering: RenderingSpec? }` +for compiler language targets (Rust/Python/Go/etc.); YAML is a Shape-B +artifact produced by `.dag` programs, never a compiler `EmissionTarget` +value. The ratified T-CI-WAD `WorkflowRuntime` below is a gunbc CI +projection mode for workflow artifact generation, not a compiler target; +the name was deliberately chosen to avoid the prior collision with +Shape-A `EmissionTarget` per PR #2749 INVARIANTS P2 finding. **Coproduct classification**: 🟡 SCAFFOLD. The initial variants are the ratified projection modes needed for the T-CI-WAD sequence. Future variants are @@ -22,16 +26,16 @@ consumer lands. ## Ratified Shape ```dag -type EmissionTarget +type WorkflowRuntime = YamlStatic | BinaryShim | PythonShim | InlineGunbc -fn project_github_actions(workflow: CIWorkflowDag, target: EmissionTarget) -> Workflow +fn project_github_actions(workflow: CIWorkflowDag, target: WorkflowRuntime) -> Workflow ``` -`EmissionTarget` is standalone. It is not a field on `CIPipeline`, +`WorkflowRuntime` is standalone. It is not a field on `CIPipeline`, `Workflow`, or a wrapper record. The selected target is a property of the projection call: @@ -46,19 +50,19 @@ workflow declaration. ## Rejected Shapes -- `CIPipeline { emission_target }`: rejected because `CIPipeline` is +- `CIPipeline { workflow_runtime }`: rejected because `CIPipeline` is gate-centric, not emission-artifact-centric; coupling projection policy to the gate list violates modeling faithfulness. - `WorkflowEmission { workflow, target }`: rejected because it introduces an implicit join and duplicate authority between workflow value and target choice. -- `extdeps.github.actions.Workflow { emission_target }`: rejected because it +- `extdeps.github.actions.Workflow { workflow_runtime }`: rejected because it puts gunbc CI policy into provider platform facts. ## Shared Ratchet Shape The focused T-CI-WAD ratchet should assert: -- `EmissionTarget` and `project_github_actions` live under `dsl/gunbc/`. +- `WorkflowRuntime` and `project_github_actions` live under `dsl/gunbc/`. - `dsl/extdeps/github/actions.dag` has no CI projection policy field. - `gunbc_ci_yml_workflow` is a derived projection binding of `project_github_actions(ci_workflow_dag, YamlStatic)`. diff --git a/docs/design-ci-workflow-emitter-dispatch.md b/docs/design-ci-workflow-emitter-dispatch.md index 3e3af0df2b9..99fb0c46f24 100644 --- a/docs/design-ci-workflow-emitter-dispatch.md +++ b/docs/design-ci-workflow-emitter-dispatch.md @@ -1,7 +1,12 @@ # CI Workflow Emitter Dispatch - Design Canvas **Status:** Design canvas for T-Workflow-As-Data FULL R3-close; consumes the -placement decision from substrate-shape comparison PR #2749. +placement decision from substrate-shape comparison PR #2749. Carrier name +**`EmissionTarget` → `WorkflowRuntime`** per PR #2749 §7.3.3 P2 +name-collision fix (Shape-A `type EmissionTarget` at +`src/v3/SELF_HOSTING.md:609` owns the original name); all references in +this canvas use `WorkflowRuntime` post-rename. Variant names + projection +function shape + Shape A/B mapping unchanged. **Authority:** Worker output for PR #2744 WI-1; substrate-shape comparison canvas PR #2749; final lane absorption proceeds through the T-WAD owner. **Scope:** Workflow emission target dispatch only. This document does not @@ -41,9 +46,11 @@ target syntax. The target is a workflow runtime shape, not a compiler policy. This canvas does not: -- implement `EmissionTarget` +- implement `WorkflowRuntime` - rewrite `.github/workflows/ci.yml` -- introduce a new `MatrixSpec`, `WorkflowRuntime`, or sibling workflow carrier +- introduce a new `MatrixSpec` or sibling workflow carrier alongside the + ratified `WorkflowRuntime` (`WorkflowRuntime` itself is the canvas's + primary substrate output and is in scope) - extend `dsl/gunbc/ci.dag` to full workflow coverage - implement affected-set lens evaluation - replace current Rust/Python/Go program emitters @@ -76,23 +83,23 @@ PR #2749 evaluated the carrier-placement question: | Option | Shape | Result | |---|---|---| -| A | Add `emission_target` to `extdeps.github.actions.Workflow` | Rejected after comparison-canvas review: violates extdeps fidelity by adding gunbc projection policy to a GitHub Actions platform carrier. | -| B | Add `emission_target` to `gunbc.ci.CIPipeline` | Rejected for this gate: `CIPipeline` is gate-centric CI intent, while the target controls projection of the GitHub Actions workflow artifact. | +| A | Add `workflow_runtime` to `extdeps.github.actions.Workflow` | Rejected after comparison-canvas review: violates extdeps fidelity by adding gunbc projection policy to a GitHub Actions platform carrier. | +| B | Add `workflow_runtime` to `gunbc.ci.CIPipeline` | Rejected for this gate: `CIPipeline` is gate-centric CI intent, while the target controls projection of the GitHub Actions workflow artifact. | | C | Add `WorkflowEmission { workflow, target }` wrapper | Rejected in wrapper form: preserves separation superficially but introduces an implicit join and duplicate authority. | -| C-refined | Put `EmissionTarget` in gunbc namespace and pass it to `project_github_actions(ci_workflow_dag, target) -> Workflow` | **Consumed here.** The target choice is invocation-time modeled data, extdeps stays provider-faithful, and the emitted `Workflow` is derived from one source. | +| C-refined | Put `WorkflowRuntime` in gunbc namespace and pass it to `project_github_actions(ci_workflow_dag, target) -> Workflow` | **Consumed here.** The target choice is invocation-time modeled data, extdeps stays provider-faithful, and the emitted `Workflow` is derived from one source. | The recommended substrate shape is a gunbc-owned enum plus projection function: ```dag -type EmissionTarget +type WorkflowRuntime = YamlStatic | BinaryShim | PythonShim -fn project_github_actions(source: CIWorkflowDag, target: EmissionTarget) -> Workflow +fn project_github_actions(source: CIWorkflowDag, target: WorkflowRuntime) -> Workflow ``` -`EmissionTarget` is a normal `.dag` sum type. It is "open" only in the same +`WorkflowRuntime` is a normal `.dag` sum type. It is "open" only in the same operational sense as other R3 extension surfaces: new variants are added to this single authority when a real consumer lands. A new variant must not create a sibling workflow carrier. @@ -108,7 +115,7 @@ project_github_actions(ci_workflow_dag, PythonShim) Migration compatibility belongs outside the authoritative projection API. A temporary fixture reader or adapter may map historical records that predate the -field to `YamlStatic`, but it must return a fully explicit `EmissionTarget` +field to `YamlStatic`, but it must return a fully explicit `WorkflowRuntime` before calling `project_github_actions`. That keeps `YamlStatic` and "omitted" from becoming two authoritative encodings of the same decision. @@ -116,7 +123,7 @@ Practice 4 coproduct receipt for the initial enum: | Receipt | Classification | Disposition | |---|---|---| -| `EmissionTarget = YamlStatic | BinaryShim | PythonShim` | 🟡 YELLOW scaffold | The variants currently encode two dimensions: artifact shape (`StaticYaml` vs thin shim) and runner realization (`compiled binary` vs emitted Python). Keep the flat enum only as the Slice 4/5 dispatch surface while there are exactly three concrete consumers. | +| `WorkflowRuntime = YamlStatic | BinaryShim | PythonShim` | 🟡 YELLOW scaffold | The variants currently encode two dimensions: artifact shape (`StaticYaml` vs thin shim) and runner realization (`compiled binary` vs emitted Python). Keep the flat enum only as the Slice 4/5 dispatch surface while there are exactly three concrete consumers. | The named dissolution trigger is the first additional shim runtime or the first need to share runner metadata across shim targets. At that point the enum must @@ -126,7 +133,7 @@ dissolve into coordinates equivalent to: type EmissionArtifactShape = StaticYaml | ThinShim type ShimRunnerKind = CompiledBinary | EmittedPython -type EmissionTarget = +type WorkflowRuntime = Static(EmissionArtifactShape) | Shim { runner: ShimRunnerKind } ``` @@ -146,7 +153,7 @@ This option is rejected. It would put the target choice on The objection is that `dsl/extdeps/github/actions.dag` describes platform constraints, not gunbc CI policy. That objection is load-bearing: -`EmissionTarget` is not a GitHub Actions provider fact and must not become a +`WorkflowRuntime` is not a GitHub Actions provider fact and must not become a field rendered into, or normalized as part of, the provider `Workflow` type. The workflow carrier already describes a provider artifact: @@ -172,7 +179,7 @@ type CIPipeline { } ``` -This is attractive because `emission_target` can be read as CI-level policy, +This is attractive because `workflow_runtime` can be read as CI-level policy, and `actions.dag` explicitly names `gunbc/ci.dag` as a consumer. It also keeps provider platform facts free of gunbc-specific emission choices. @@ -187,7 +194,7 @@ It names `CIGate`s and their structural gate sources. It does not own: - step rendering - provider-specific syntax obligations -Putting `emission_target` on `CIPipeline` would require the emitter to join a +Putting `workflow_runtime` on `CIPipeline` would require the emitter to join a pipeline declaration to a separate `Workflow` declaration. That join becomes a new coherence surface: which pipeline emits which workflow, which value wins if multiple pipelines reference the same workflow, and how a target-specific shim @@ -205,7 +212,7 @@ A wrapper shape would be: ```dag type WorkflowEmission { workflow: Workflow - target: EmissionTarget + target: WorkflowRuntime } ``` @@ -248,10 +255,10 @@ This keeps all three layers distinct: | Layer | Authority | |---|---| | CI semantics | `gunbc.ci` workflow/gate graph | -| Emission choice | invocation-time `EmissionTarget` argument | +| Emission choice | invocation-time `WorkflowRuntime` argument | | Provider artifact | derived `extdeps.github.actions.Workflow` | -The cost of adding a target remains bounded: add one `EmissionTarget` variant +The cost of adding a target remains bounded: add one `WorkflowRuntime` variant and one projection consumer. The extdeps carrier remains unchanged. ### 3.2 Carrier Reuse Audit @@ -303,7 +310,7 @@ it for Slice 4 or Slice 5. The workflow emitter has one structural responsibility: ``` -(CI workflow graph, EmissionTarget, GitHub Actions platform facts, +(CI workflow graph, WorkflowRuntime, GitHub Actions platform facts, target language/runtime facts) -> emitted workflow artifact(s) ``` @@ -323,7 +330,7 @@ The dispatch steps are: Pseudocode: ```dag -fn emit_workflow(source: CIWorkflowDag, target: EmissionTarget) -> WorkflowEmissionResult = +fn emit_workflow(source: CIWorkflowDag, target: WorkflowRuntime) -> WorkflowEmissionResult = let workflow = project_github_actions(source, target) match target { YamlStatic => @@ -465,7 +472,7 @@ orchestrator. GitHub Actions becomes a minimal host that invokes gunbc's workflow interpreter directly. This target is intentionally sketched only. It should not block Slice 4 or -Slice 5. It is not part of the initial `EmissionTarget` substrate enum. Its +Slice 5. It is not part of the initial `WorkflowRuntime` substrate enum. Its value is architectural: the target surface can later express "workflow stays inside gunbc runtime" without inventing a second workflow model. @@ -567,7 +574,7 @@ and repository state: This yields two TestClaim shapes: ```dag -data workflow_emission_target_toggle_proven: TestClaim = ... +data workflow_runtime_toggle_proven: TestClaim = ... data workflow_target_semantics_equivalent: TestClaim = ... ``` @@ -616,7 +623,7 @@ This canvas keeps Slice 4 and Slice 5 small: | Slice | Dependency | |---|---| -| Slice 4: projection + `YamlStatic` | gunbc-owned `EmissionTarget` plus `project_github_actions(..., YamlStatic)` | +| Slice 4: projection + `YamlStatic` | gunbc-owned `WorkflowRuntime` plus `project_github_actions(..., YamlStatic)` | | Slice 5: `BinaryShim` | Slice 4 target dispatch plus compiled runner surface | | Slice 7: affected-set CI | Slice 5 plus affected-set lens | | Slice 8: `ci.yml` deletion | Slice 4-7 accepted | @@ -633,13 +640,13 @@ is shaped to consume affected-set receipts when they become available. ### Phase A: Ratify Shape - Land this canvas. -- Ratify `EmissionTarget` placement in gunbc namespace as the projection +- Ratify `WorkflowRuntime` placement in gunbc namespace as the projection argument to `project_github_actions`. - Ratify that authoritative projection calls require an explicit target. ### Phase B: Static Parity -- Add `EmissionTarget` to gunbc CI/emission substrate, not to extdeps. +- Add `WorkflowRuntime` to gunbc CI/emission substrate, not to extdeps. - Model current CI in `dsl/gunbc/ci.dag` as the semantic source. - Emit `.github/workflows/ci.yml` from `project_github_actions(ci_workflow_dag, YamlStatic)`. @@ -677,7 +684,7 @@ Recommended ratchets: | Ratchet | Purpose | |---|---| -| `workflow_emission_target_consumed` | `EmissionTarget` is read by the workflow projection/emitter. | +| `workflow_runtime_consumed` | `WorkflowRuntime` is read by the workflow projection/emitter. | | `workflow_yaml_static_fresh` | emitted YAML matches checked-in artifact while static artifact remains checked in. | | `workflow_binary_shim_is_thin` | shim YAML contains only bootstrap/checkout/setup/invoke steps. | | `workflow_no_path_regex_policy` | no durable CI selection policy remains in YAML path regexes after affected-set lands. | @@ -693,7 +700,7 @@ The following questions should be answered by ratification, not by worker implementation: 1. Which non-authoritative fixture reader or migration adapter should translate - legacy records that predate `EmissionTarget`? + legacy records that predate `WorkflowRuntime`? 2. Should `BinaryShim` first run jobs in-process, or should it emit dynamic matrix JSON for GitHub fanout? 3. What is the minimal verifier for GitHub Actions YAML in CI before a full @@ -712,7 +719,7 @@ This canvas recommends: This design is ready for downstream implementation when: -- `EmissionTarget` lives in gunbc namespace as a projection argument, with +- `WorkflowRuntime` lives in gunbc namespace as a projection argument, with explicit target required at authoritative projection call sites. - `YamlStatic`, `BinaryShim`, and `PythonShim` have explicit semantics and acceptance contracts. diff --git a/docs/design-ci-workflow-substrate-shape-2026-05-12.md b/docs/design-ci-workflow-substrate-shape-2026-05-12.md index 1cae6418d21..261065d5271 100644 --- a/docs/design-ci-workflow-substrate-shape-2026-05-12.md +++ b/docs/design-ci-workflow-substrate-shape-2026-05-12.md @@ -3,9 +3,9 @@ **Status:** **RATIFIED** by Director (zesty-bear-812) 2026-05-12T07:04:18Z per PR #2749 comment (Director session msg_4f7f536d to PM deep-wolf-155). Option **(c-refined)** is the ratified substrate-shape for gate #56 -`ci_workflow_modeled_as_dag` under T-WAD FULL R3-close: `EmissionTarget` +`ci_workflow_modeled_as_dag` under T-WAD FULL R3-close: `WorkflowRuntime` sum (🟡 YELLOW per §7.3.1) in gunbc namespace; `project_github_actions: -(CIWorkflowDag, EmissionTarget) → Workflow` in gunbc; `extdeps.github.actions.Workflow` +(CIWorkflowDag, WorkflowRuntime) → Workflow` in gunbc; `extdeps.github.actions.Workflow` unmodified; `Workflow` values strictly derived as projection outputs. Ratification asks #1, #2, #3, #4, #5 dispositioned (see §9). @@ -105,7 +105,7 @@ implies a deletion that the diff does not perform. ### Option (b) — Provider-specific concrete carriers (WI-1 canvas, PR #2746) **Carrier surface**: `dsl/extdeps/github/actions.dag`'s existing `Workflow` -gains an `emission_target: EmissionTarget?` field; `EmissionTarget` is a +gains an `workflow_runtime: WorkflowRuntime?` field; `WorkflowRuntime` is a new sum type `= YamlStatic | BinaryShim | PythonShim | InlineGunbc`. Same canvas (PR #2746 `docs/design-ci-workflow-emitter-dispatch.md` §3) rejects sibling `WorkflowEmission { workflow, target }` wrappers (Option C in that @@ -113,7 +113,7 @@ canvas) and rejects placing the field on `gunbc.ci.CIPipeline` (Option B in that canvas). **Authority claim**: the existing `Workflow` carrier (Job/Step nested) is the -modeled authority for the workflow artifact; `emission_target` selects which +modeled authority for the workflow artifact; `workflow_runtime` selects which projection shape (YamlStatic / BinaryShim / PythonShim / InlineGunbc) the single-emitter renders from that one `Workflow` value. @@ -124,10 +124,10 @@ without an explicit node/edge prerequisite graph. ### Option (c) — Hybrid: provider-neutral semantic source + actions.dag-composing projection layer -> **⚠ SUPERSEDED.** This option-(c) framing places `emission_target` on +> **⚠ SUPERSEDED.** This option-(c) framing places `workflow_runtime` on > `extdeps.github.actions.Workflow` and refers to a hand-declared > "pinned `Workflow` value" the emitter validates against. Both are -> retracted: §7 (INVARIANTS P2/P1) moves `EmissionTarget` to gunbc +> retracted: §7 (INVARIANTS P2/P1) moves `WorkflowRuntime` to gunbc > namespace as a projection-function parameter, and §8 retracts the > pinned-`Workflow`-as-modeled-authority framing. The ratified > substrate-shape is **(c-refined)** in §7.3 / §9. The text below is @@ -135,14 +135,14 @@ without an explicit node/edge prerequisite graph. > read §7.3 + §7.4 + §8 + §9. **Carrier surface**: option (a)'s `CIWorkflowDag` retained as semantic -gate-dependency authority **plus** option (b)'s `emission_target: EmissionTarget?` +gate-dependency authority **plus** option (b)'s `workflow_runtime: WorkflowRuntime?` retained on `extdeps.github.actions.Workflow`. The two carriers occupy **different concept layers**: - `gunbc.ci.CIWorkflowDag` — semantic source: which gates exist + how they depend (provider-neutral; usable by any CI provider or by `dag run` directly) - `extdeps.github.actions.Workflow` — provider transport: GitHub Actions - workflow artifact, with `emission_target` selecting which shape the + workflow artifact, with `workflow_runtime` selecting which shape the single-emitter renders **Authority claim**: each carrier is single-authority **for its layer**. @@ -219,7 +219,7 @@ Slice 8 (ci.yml deletion) cannot ground without a workflow-artifact carrier. ### §2.2 Option (b) evaluation -**Cost-of-change**: low for new emission targets (one `EmissionTarget` +**Cost-of-change**: low for new emission targets (one `WorkflowRuntime` variant + one emitter consumer; explicitly stated in PR #2746 canvas §3.1). **High for new CI providers** — option (b) is GitHub-Actions-specific by construction; another provider needs its own concrete carriers (GitLab CI's @@ -255,7 +255,7 @@ property too (else any new CI provider re-shadows the same fact). ### §2.3 Option (c) evaluation > **⚠ SUPERSEDED.** Evaluation below assumes option (c) as authored (with -> `emission_target` on `extdeps.github.actions.Workflow` and a hand- +> `workflow_runtime` on `extdeps.github.actions.Workflow` and a hand- > declared "pinned `Workflow`" validation target). Both elements are > retracted in §7 (P2/P1) and §8 (single-authority). Ratified shape is > (c-refined) in §7.3 + §7.4 + §9. Specifically: the "emission validates @@ -265,7 +265,7 @@ property too (else any new CI provider re-shadows the same fact). > the sole authority. Section preserved as canvas-evolution record. **Cost-of-change**: low across both axes — new provider = new projection -function (semantic source unchanged); new emission target = one `EmissionTarget` +function (semantic source unchanged); new emission target = one `WorkflowRuntime` variant + emitter consumer (semantic source unchanged). The two cost-of-change axes are **decoupled** because the carriers live on different concept layers. @@ -302,7 +302,7 @@ ci.yml under the YamlStatic emission target. |---|---|---|---| | Gate-dependency authority | ✓ `CIWorkflowDag` | ✗ implicit in `Job.needs` strings | ✓ `CIWorkflowDag` | | Workflow-artifact authority | ✗ unmodeled | ✓ `Workflow` | ✓ `Workflow` | -| Emission-target selector | ✗ build-system implicit | ✓ `emission_target` field | ✓ `emission_target` field | +| Emission-target selector | ✗ build-system implicit | ✓ `workflow_runtime` field | ✓ `workflow_runtime` field | | Single-authority per layer | ✗ dual-authority unresolved | ✓ (transport only) | ✓ both layers | | Cost-of-change: new provider | low | high (provider-specific) | low | | Cost-of-change: new target | high (no selector) | low | low | @@ -315,13 +315,13 @@ ci.yml under the YamlStatic emission target. ## §3. WI-1 canvas (PR #2746) assumption-breakage under (c) > **⚠ SUPERSEDED.** This section analyzed WI-1 assumptions under option -> (c) as originally authored — keeping `emission_target` on +> (c) as originally authored — keeping `workflow_runtime` on > `extdeps.github.actions.Workflow`. Under the ratified (c-refined) in -> §7.3, the substrate decision on `EmissionTarget` placement is **a +> §7.3, the substrate decision on `WorkflowRuntime` placement is **a > substantive retraction**, not a framing-narrowing: PR #2746's "field -> on `Workflow`" placement is withdrawn in favor of `EmissionTarget` in +> on `Workflow`" placement is withdrawn in favor of `WorkflowRuntime` in > gunbc namespace as a projection-function parameter. Specifically: -> the `"emission_target` field on `Workflow` is correct placement" +> the `"workflow_runtime` field on `Workflow` is correct placement" > bullet below is OVERTURNED by §7.1; the "Workflow data chooses its > emission target" framing is OVERTURNED by §7.6; the "merge with a > one-paragraph clarifying note" disposition is OVERTURNED by §7.5 ask @@ -337,10 +337,10 @@ workflow as a whole — `CIWorkflowDag` owns gate-dependency. The assumptions that hold: -- **`emission_target` field on `Workflow` is correct placement** — per PR #2746 §3, +- **`workflow_runtime` field on `Workflow` is correct placement** — per PR #2746 §3, the field selects artifact projection shape, which is a property of the transport artifact. (c) does not move this field. -- **`EmissionTarget` sum is correct shape** — `YamlStatic | BinaryShim | PythonShim | InlineGunbc` +- **`WorkflowRuntime` sum is correct shape** — `YamlStatic | BinaryShim | PythonShim | InlineGunbc` describes projection-target choices on the artifact; orthogonal to the semantic gate-dependency DAG. - **Optional field + `none` → `YamlStatic` migration normalization is correct** — @@ -348,7 +348,7 @@ The assumptions that hold: - **PR #2746 §3 rejection of sibling `WorkflowEmission { workflow, target }`** — remains correct under (c) for the same reason (would create implicit join between the wrapper and the projection function). -- **PR #2746 §3 rejection of `emission_target` on `CIPipeline`** — remains +- **PR #2746 §3 rejection of `workflow_runtime` on `CIPipeline`** — remains correct: `CIPipeline` is gate-centric, not artifact-centric. The assumptions that break: @@ -363,7 +363,7 @@ The assumptions that break: projection function chooses (or is parameterized by) the emission target; the declared `Workflow` is the validated projection result for one emission target. Multi-target emission means multiple projection function - invocations, not multiple `emission_target` fields on one `Workflow`. + invocations, not multiple `workflow_runtime` fields on one `Workflow`. - **Slice 4/5 acceptance contract framing** — PR #2746 §4 (per body summary) describes emitter implementations consuming `Workflow`. Under (c), the implementations consume the **projection function output** for the @@ -379,7 +379,7 @@ authority is the part that needs narrowing to "workflow-artifact authority **Recommendation for PR #2746 disposition**: merge with a one-paragraph clarifying note at §0 (or in PR body) tying `Workflow` to the transport layer and pointing to this canvas for the gate-dependency authority. The -`EmissionTarget` substrate decision itself stands; only the surrounding +`WorkflowRuntime` substrate decision itself stands; only the surrounding framing narrows. --- @@ -389,7 +389,7 @@ framing narrows. > **⚠ SUPERSEDED in part.** Sequencing structure (S0→S1→S2∥S3→S4) and > the load-bearing role of S1 (projection function) REMAIN VALID under > (c-refined). What changes: any S0 / S1 description that places -> `EmissionTarget?` on `Workflow` or refers to a hand-declared +> `WorkflowRuntime?` on `Workflow` or refers to a hand-declared > `Workflow` validation target is overturned by §7/§8. WI-2 (PR #2745) > re-brief scope is also LARGER than §5.4 implied — see §9 ratification > ask #4. For current sequencing read this section IN COMBINATION with @@ -398,7 +398,7 @@ framing narrows. The Director-ratified gate-additions (per PM relay `msg_93e14076`): -- `workflow_emission_target_toggle_proven` (substrate-shape) — Slice 4 + Slice 5 +- `workflow_runtime_toggle_proven` (substrate-shape) — Slice 4 + Slice 5 - `ci_yml_dissolved` (state-check) — Slice 8 - `ci_uses_affected_set_selection` (state-check) — Slice 7 (Verification Mgr lane) - `test_cost_dimension_landed` (substrate-shape + state-check) — Slice 6 (Debt-Paydown Mgr lane) @@ -422,23 +422,23 @@ Under option (c) the lane-absorbed gate-set sequences as: **S1 — Projection function (new sub-slice; not yet in Director-ratified gate-set)** - New brief: `project_github_actions: CIWorkflowDag → Workflow` lives at `dsl/gunbc/ci_emission.dag` (or similar) and consumes `CIWorkflowDag` from - `gunbc.ci` + emits a `Workflow` value parameterized by `EmissionTarget`. + `gunbc.ci` + emits a `Workflow` value parameterized by `WorkflowRuntime`. - This sub-slice is the load-bearing substrate fact that ties option (a) + option (b) carriers into option (c)'s hybrid. **Without it, (c) reduces to dual-authority (a)+(b) coexistence.** - Whether the projection function is itself a new §1.8 gate or part of - `workflow_emission_target_toggle_proven` acceptance is a Director call; + `workflow_runtime_toggle_proven` acceptance is a Director call; this canvas surfaces the question without selecting. **S2 — Slice 4 (YamlStatic emitter)** - Implements emitter consuming `project_github_actions(ci_workflow_dag, YamlStatic)` → ci.yml-equivalent artifact - Acceptance: emitter output structurally equivalent to current `.github/workflows/ci.yml` (workflow-semantics, not byte equality — per PR #2746 §2) -- Half of `workflow_emission_target_toggle_proven` (one target proven) +- Half of `workflow_runtime_toggle_proven` (one target proven) **S3 — Slice 5 (BinaryShim emitter)** - Implements emitter consuming `project_github_actions(ci_workflow_dag, BinaryShim)` → thin YAML shim invoking compiled gunbc CI binary - Acceptance: BinaryShim output executes the same `CIWorkflowDag` gates that YamlStatic would -- Completes `workflow_emission_target_toggle_proven` (toggle proven across two targets from same `CIWorkflowDag`) +- Completes `workflow_runtime_toggle_proven` (toggle proven across two targets from same `CIWorkflowDag`) **S4 — Slice 8 (ci.yml dissolution)** - Delete hand-authored `.github/workflows/ci.yml`; replace with emitted artifact (Slice 4 YamlStatic output) or BinaryShim (Slice 5 output) @@ -460,7 +460,7 @@ S6 is fully orthogonal. Slice 4 ∥ Slice 5 can dispatch in parallel once S1 ## §5. Recommendation > **⚠ SUPERSEDED.** Section recommends "option (c)" with -> `Workflow.emission_target` on `dsl/extdeps/github/actions.dag`, asks +> `Workflow.workflow_runtime` on `dsl/extdeps/github/actions.dag`, asks > for PR #2746 merge with "no substrate retraction" + a pinned > `Workflow` validation target. All three positions are OVERTURNED: > §7 (P2/P1) retracts the extdeps placement; §8 retracts the pinned- @@ -471,7 +471,7 @@ S6 is fully orthogonal. Slice 4 ∥ Slice 5 can dispatch in parallel once S1 **Adopt option (c) — hybrid: provider-neutral semantic source (`CIWorkflowDag` -at `dsl/gunbc/ci.dag`) + actions.dag-composing projection layer (`Workflow.emission_target` +at `dsl/gunbc/ci.dag`) + actions.dag-composing projection layer (`Workflow.workflow_runtime` at `dsl/extdeps/github/actions.dag`) + projection function (`project_github_actions: CIWorkflowDag → Workflow`).** @@ -485,7 +485,7 @@ CIWorkflowDag → Workflow`).** scope doc §1 elevation framing). 3. Only (c) preserves both already-authored PRs' substrate contributions: PR #2736's `CIWorkflowDag` retained as semantic source; PR #2746's - `EmissionTarget` retained as transport-artifact selector. The + `WorkflowRuntime` retained as transport-artifact selector. The reconciliation is a framing narrowing on PR #2746 (not a substrate retraction) + a PR body correction on PR #2736 (single-authority claim narrowed to the gate-dependency layer). @@ -518,7 +518,7 @@ CIWorkflowDag → Workflow`).** pinned `Workflow` value, not a bare `gunbc_ci_yml_workflow` value at `gunbc.ci`. Or hold PR #2745 closure pending S1 brief. 5. **Decide whether S1 (projection function) is a new §1.8 gate** or part - of `workflow_emission_target_toggle_proven` acceptance. + of `workflow_runtime_toggle_proven` acceptance. **If Director rejects (c)** in favor of (a) or (b): the dual-authority condition at HEAD remains and needs separate retraction work. Under (a), @@ -537,10 +537,10 @@ Both rejections cost more substrate churn than (c) ratification. 1. **S1 projection function placement**: `dsl/gunbc/ci_emission.dag`? `dsl/gunbc/ci.dag` itself? A new module? Director-tier concept-layering call. -2. **Projection function signature**: take `EmissionTarget` as parameter, or +2. **Projection function signature**: take `WorkflowRuntime` as parameter, or read it from the input `CIWorkflowDag` (would require adding the field there, contradicting M9)? **Resolved in §7.3 (c-refined)**: signature is - `(CIWorkflowDag, EmissionTarget) → Workflow`, with `EmissionTarget` as + `(CIWorkflowDag, WorkflowRuntime) → Workflow`, with `WorkflowRuntime` as parameter — NOT a field on any carrier. The "PR #2746 places the field on `Workflow`" framing in the pre-§7 reasoning below is **overturned**: under (c-refined), `extdeps.github.actions.Workflow` is unmodified @@ -567,7 +567,7 @@ Both rejections cost more substrate churn than (c) ratification. 2026-05-12T06:44:13Z (comment c#4427988541), relayed by PM `msg_31090356`: -> Adding `EmissionTarget` to `dsl/extdeps/github/actions.dag` puts a gunbc +> Adding `WorkflowRuntime` to `dsl/extdeps/github/actions.dag` puts a gunbc > emission/runtime choice into the GitHub Actions platform model, violating > extdeps fidelity and INVARIANTS P1. @@ -583,7 +583,7 @@ per Director, 2026-05-12): The BLOCKING finding is **structurally correct**. The extdeps fidelity boundary is explicit at the file header: `extdeps.github.actions` models **platform facts** (what the runtime consumes); `gunbc/ci.dag` models **CI -logic** (what gunbc emits/dispatches). `EmissionTarget` (`YamlStatic | +logic** (what gunbc emits/dispatches). `WorkflowRuntime` (`YamlStatic | BinaryShim | PythonShim | InlineGunbc`) is a gunbc emission-policy fact — the runtime does not consume "emission target"; it consumes whatever YAML or shim ends up at `.github/workflows/ci.yml`. The selection between @@ -602,7 +602,7 @@ extdeps and gunbc. Subsequent canvas references to single-authority cite Per `INVARIANTS.md` P2 (Boundary Discipline): emission-policy authority belongs to `gunbc/ci.dag` (single authority), **not** `extdeps.github.actions` -(which holds platform authority). Placing `emission_target` on +(which holds platform authority). Placing `workflow_runtime` on `extdeps.github.actions.Workflow` creates a dual-authority condition where the extdeps file holds a gunbc-policy fact alongside its platform facts. Per P1 (Modeling Faithfulness): the extdeps carrier would no longer @@ -631,24 +631,24 @@ runtime/emitter boundary. This addendum **disqualifies** §1 option (b) AS-AUTHORED and **partially invalidates** §1 option (c) as authored above: -- **Option (b) — `EmissionTarget` on `actions.Workflow`**: DISQUALIFIED per +- **Option (b) — `WorkflowRuntime` on `actions.Workflow`**: DISQUALIFIED per §7.1. The substrate-decision in PR #2746 stands as a SHAPE (sum type needed; sibling wrapper still rejected for the same join-cost reason), but the PLACEMENT moves out of extdeps. -- **Option (c) — hybrid retaining `Workflow.emission_target`**: PARTIALLY +- **Option (c) — hybrid retaining `Workflow.workflow_runtime`**: PARTIALLY INVALIDATED. The two-layer concept-layering argument (semantic source at `gunbc.ci.CIWorkflowDag` + transport at `extdeps.github.actions.Workflow`) - REMAINS CORRECT. The `EmissionTarget` placement on the transport carrier + REMAINS CORRECT. The `WorkflowRuntime` placement on the transport carrier is what fails P1. The §2 four-axis comparison conclusions REMAIN VALID for the layering question: only a hybrid satisfies single-authority per layer + concept -layering simultaneously. What changes is **where `EmissionTarget` lives** +layering simultaneously. What changes is **where `WorkflowRuntime` lives** within the hybrid. ### §7.3 Refined option (c) — "(c-refined)" -`EmissionTarget` lives in **`gunbc/ci.dag`** (gunbc namespace; emission- +`WorkflowRuntime` lives in **`gunbc/ci.dag`** (gunbc namespace; emission- policy authority) as a sum type. It is **not** a field on any extdeps carrier. The projection function takes it as a parameter: @@ -661,23 +661,23 @@ carrier. The projection function takes it as a parameter: // (b) consumer needs single-dimension pattern-match, (c) Slice 4/5 // surfaces unpredicted axis). Likely dissolution path: Pattern 4 // (Dimensional) over { target_language, requires_shim, runtime_executes }. -type EmissionTarget +type WorkflowRuntime = YamlStatic | BinaryShim | PythonShim | InlineGunbc -// project_github_actions: CIWorkflowDag x EmissionTarget -> Workflow +// project_github_actions: CIWorkflowDag x WorkflowRuntime -> Workflow // Pure structural projection (per docs/design-emission-model.md): // reads gate-dependency DAG + target selector, emits actions.Workflow // value that the emitter renders to the chosen artifact shape. fn project_github_actions( ci_workflow_dag: CIWorkflowDag, - target: EmissionTarget, + target: WorkflowRuntime, ) -> Workflow ``` -`extdeps.github.actions.Workflow` is **unmodified** — no `emission_target` +`extdeps.github.actions.Workflow` is **unmodified** — no `workflow_runtime` field added. The platform model stays bounded to platform facts. The `Workflow` value WI-2 was authoring at PR #2745 (the structural mirror @@ -690,11 +690,11 @@ a binding to the derived result; there is no separate hand-declared WI-2 substrate work to gunbc namespace under (c-refined) AND tightens it to single-authority projection-output binding. -#### §7.3.1 `EmissionTarget` coproduct dissolution classification +#### §7.3.1 `WorkflowRuntime` coproduct dissolution classification Per `docs/modeling-discipline.md` Practice 4: every coproduct with N ≥ 2 variants must carry a 🟢/🟡/🔴 classification + ledger entry (GREEN) or -named trigger (YELLOW). `EmissionTarget` has four variants and was +named trigger (YELLOW). `WorkflowRuntime` has four variants and was proposed without classification in PR #2746 + this canvas's §7.3 draft — the canvas cannot ratify an unclassified sum. @@ -706,7 +706,7 @@ the canvas cannot ratify an unclassified sum. to distribute to per-variant consumers. The variant tag itself selects which projection-function branch runs. Pattern 1 does not apply. 2. **Variant-is-data** — same structural shape (all empty), different - labels. Could potentially promote to `EmissionTarget { name: String }`, + labels. Could potentially promote to `WorkflowRuntime { name: String }`, but the modeling-discipline.md guardrail rules this out: "only valid when the label space is closed and enumerable, not when it's free-form string." A free-form `name: String` would dilute closed-set discipline. @@ -740,7 +740,7 @@ dissolution that's likely correct but premature. **Named dissolution trigger** (per YELLOW requirements): dissolve when EITHER (a) a fifth target lands that does not fit cleanly into the existing four-way axis (e.g., `RustShim`, `GoShim` would force the shim+ -language dimensions to be explicit), OR (b) a consumer of `EmissionTarget` +language dimensions to be explicit), OR (b) a consumer of `WorkflowRuntime` emerges that needs to pattern-match on a single dimension rather than the full variant tag (e.g., an emitter that wants "render YAML regardless of shim presence"), OR (c) Slice 4/Slice 5 implementation surfaces an axis @@ -752,7 +752,7 @@ to discover the right axes from real consumer pressure. into the `.dag` declaration (which does not yet exist). The implementing PR (WI-2 re-brief per §5/§7.5 ask #4) MUST carry the same classification + trigger as a `// 🟡 YELLOW (scaffold)` comment on the `type -EmissionTarget` declaration when it lands, citing this canvas's §7.3.1. +WorkflowRuntime` declaration when it lands, citing this canvas's §7.3.1. #### §7.3.2 Shape A / Shape B clarification — what "emission target" names @@ -776,12 +776,12 @@ shape concern. Per `THESIS.md:215-216`: A YAML file is a Shape B artifact. The gunbc compiler does NOT emit YAML for `.github/workflows/ci.yml`; a `.dag` user program (the projection's realization layer) walks the typed `Workflow` value and renders YAML. -Naming a `YamlStatic` variant on `EmissionTarget` invites the misread +Naming a `YamlStatic` variant on `WorkflowRuntime` invites the misread that gunbc has a "YAML emission target" parallel to Rust/Python/etc. **Resolution** (clarification, not retraction): -`EmissionTarget` does NOT name a compiler emission target. It names a +`WorkflowRuntime` does NOT name a compiler emission target. It names a **realization mode** — which downstream path the projected `Workflow` value flows through to become the actual CI artifact at `.github/workflows/ci.yml`. The mapping is: @@ -793,7 +793,7 @@ value flows through to become the actual CI artifact at | `PythonShim` | gunbc compiles to Python (Shape A) + thin YAML shim (Shape B) invokes it | mixed Shape A + Shape B | THESIS:215 + :216 | | `InlineGunbc` | gunbc runtime orchestrates gates directly | Shape A (gunbc as host) | THESIS:215 | -The substrate decision is correct: `EmissionTarget` selects which +The substrate decision is correct: `WorkflowRuntime` selects which realization path; the projection function output is a typed `Workflow` regardless of path. **What this clarification fixes is reader expectations**: the variants are not parallel compiler emission targets; they are @@ -803,10 +803,10 @@ realize into the final artifact. **Naming consideration**: a name like `WorkflowRealizationMode` or `ArtifactRealizationKind` would carry less Shape-A baggage than -`EmissionTarget`. PR #2746 + Director ratification + ongoing downstream -worker briefs use `EmissionTarget`; renaming now would force re- +`WorkflowRuntime`. PR #2746 + Director ratification + ongoing downstream +worker briefs use `WorkflowRuntime`; renaming now would force re- ratification work without a corresponding substrate change. Recommendation: -**keep the `EmissionTarget` name; document the Shape A/B realization +**keep the `WorkflowRuntime` name; document the Shape A/B realization mapping at the declaration site** (when the type lands in `.dag` per §7.5 ask #4) so consumers reading the carrier understand it as a realization-mode selector, not a "gunbc emits YAML" claim. @@ -814,10 +814,53 @@ realization-mode selector, not a "gunbc emits YAML" claim. **No substrate retraction**: the four variants, the 🟡 YELLOW classification, the three-condition dissolution trigger, the gunbc- namespace placement, and the parametric projection-function signature -all stand. Only the implicit framing that `EmissionTarget` named +all stand. Only the implicit framing that `WorkflowRuntime` named "compiler emission targets" is corrected — it names realization modes, several of which are Shape B (user-program rendering). +#### §7.3.3 Carrier rename — `EmissionTarget` → `WorkflowRuntime` + +**Trigger** (operator BLOCKING on the unmerged PR #2749 branch at :666, +2026-05-12T09:26:50Z; landed in main via cascade PR #2756): + +> Keeping the `EmissionTarget` name for CI realization modes creates a +> second incompatible authority beside `src/v3/SELF_HOSTING.md`'s +> Shape-A `EmissionTarget`, violating INVARIANTS P2 single-authority. + +**Verified**: `src/v3/SELF_HOSTING.md:609` declares the canonical +Shape-A compiler-emission carrier: + +```dag +type EmissionTarget { + language: LanguageSpec // what's valid (required) + rendering: RenderingSpec? // how to format (optional) +} +``` + +That carrier is the authoritative `EmissionTarget` for Shape-A compiler +emission targets (Rust/Python/Go/etc.). This canvas's original §7.3 +sketch used the same name for a fundamentally different carrier (CI +realization-mode selector) — literal name collision. Per +`INVARIANTS.md` P2 (Boundary Discipline), every fact lives in exactly +one authoritative place, named once. §7.3.2's "keep the name, document +the mapping" disposition is insufficient: no amount of framing- +narrowing keeps a name available when another carrier already owns it. + +**Rename**: `EmissionTarget` → `WorkflowRuntime` throughout the canvas, +with cascade across PR #2744 (briefs), PR #2745 (WI-2 impl), PR #2750 +(ledger gate-IDs), and PR #2746-merged content (this PR #2756). Variant +names + 23-site migration scope + 🟡 YELLOW classification + +three-condition dissolution trigger + projection-function signature + +gunbc-namespace placement all stand; only the sum-type name changed. + +**Director re-ratification anchor**: Director affirmed the rename via +PM relay `msg_5dd7d82d` 2026-05-12: "Director ratification: AFFIRM the +cascade-rename. gunbc-namespace EmissionTarget → WorkflowRuntime. +SELF_HOSTING.md:609 stays untouched (pre-existing authority; Shape-A +compiler emitter-composition). All 4 in-flight rename commits stand." +This supersedes the original `msg_4f7f536d` ratification at the +sum-type-name level only; all other ratified elements unchanged. + ### §7.4 Comparison table revised Replacing the §2.4 row "Workflow-artifact authority": @@ -826,14 +869,14 @@ Replacing the §2.4 row "Workflow-artifact authority": |---|---|---|---|---| | Gate-dependency authority | ✓ `CIWorkflowDag` | ✗ implicit `Job.needs` | ✓ `CIWorkflowDag` | ✓ `CIWorkflowDag` | | Workflow-artifact authority | ✗ unmodeled | ✓ `Workflow` (platform-coupled emission-policy) | ✓ `Workflow` (same) | ✓ `Workflow` (platform-pure) + projection-fn in gunbc | -| Emission-target carrier | ✗ build-system implicit | ✗ on extdeps (P1 violation) | ✗ on extdeps (P1 violation) | ✓ `gunbc.ci.EmissionTarget` (gunbc-namespace) | +| Emission-target carrier | ✗ build-system implicit | ✗ on extdeps (P1 violation) | ✗ on extdeps (P1 violation) | ✓ `gunbc.ci.WorkflowRuntime` (gunbc-namespace) | | INVARIANTS P2 (Boundary Discipline / single authority) | ✗ dual-authority unresolved | ✗ extdeps holds gunbc-policy fact | ✗ extdeps holds gunbc-policy fact | ✓ each authority single-sourced | | Extdeps fidelity (file header :1-12) | n/a | ✗ violated | ✗ violated | ✓ preserved | | Sufficient for FULL R3-close | no | no | no (P1) | **yes** | ### §7.5 Revised recommendation -**Adopt option (c-refined): `EmissionTarget` lives in `gunbc/ci.dag` as a +**Adopt option (c-refined): `WorkflowRuntime` lives in `gunbc/ci.dag` as a parameter to `project_github_actions(ci_workflow_dag, target) → Workflow`; `extdeps.github.actions.Workflow` stays unmodified; the only `Workflow` values in modeled authority are bindings to the projection function's @@ -844,9 +887,9 @@ No separately-declared `Workflow` value exists as modeled authority.** 1. **Ratify (c-refined)** as gate #56 substrate-shape under T-CI-WAD program-tag. 2. **PR #2746 disposition changes from "merge with framing-narrowing - addendum" to "substantive substrate retraction"**: the `EmissionTarget? + addendum" to "substantive substrate retraction"**: the `WorkflowRuntime? on actions.Workflow` field placement is withdrawn. PR #2746 canvas - §3.1 reasoning narrows to "`EmissionTarget` substrate-decision + §3.1 reasoning narrows to "`WorkflowRuntime` substrate-decision (sum-type, not sibling-wrapper) is correct; placement moves from extdeps to gunbc namespace per INVARIANTS P2 (boundary discipline) + P1 (extdeps-carrier modeling faithfulness)". Suggest PR #2746 @@ -875,12 +918,12 @@ shape. ### §7.6 Why (c-refined) is not PM's "(b)" or "(c)" in `msg_31090356` PM relay `msg_31090356` proposed: -- PM (b) — `EmissionTarget` on `gunbc.ci.CIPipeline` +- PM (b) — `WorkflowRuntime` on `gunbc.ci.CIPipeline` - PM (c) — Hybrid wrapper `WorkflowEmission { workflow: actions.Workflow, target }` in gunbc namespace (c-refined) is neither: -- **vs PM (b)**: putting `EmissionTarget` on `CIPipeline` couples emission- +- **vs PM (b)**: putting `WorkflowRuntime` on `CIPipeline` couples emission- policy to the gate-list carrier. The same join-cost problem WI-1 canvas §3 Option B already flagged applies — `CIPipeline` is gate-centric, not emission-artifact-centric. PM (b) trades one P1 problem for a different @@ -898,9 +941,9 @@ projection; the projection function reads target spec + source; there is no carrier-level "this Workflow is targeted at X" fact because emission choice is a property of the emit-call, not a property of the value. -The `EmissionTarget` sum-type itself lives in gunbc namespace (as a +The `WorkflowRuntime` sum-type itself lives in gunbc namespace (as a named type usable anywhere gunbc CI logic needs to reference emission -targets), but no `EmissionTarget` **field** lives on any existing +targets), but no `WorkflowRuntime` **field** lives on any existing carrier. This eliminates both P1 (no extdeps modification) and M9 (no gate-centric carrier extension). @@ -952,7 +995,7 @@ ask #5, §7.3, §7.5): The §2 / §7 layering argument is unchanged: gate-dependency lives at `gunbc.ci.CIWorkflowDag`; transport platform model lives at `extdeps.github.actions.Workflow`; emission policy -(`EmissionTarget` + `project_github_actions`) lives in gunbc namespace; +(`WorkflowRuntime` + `project_github_actions`) lives in gunbc namespace; the workflow artifact is derived by the projection function from a single source. @@ -975,9 +1018,9 @@ Director (zesty-bear-812) ratified option **(c-refined)** per PR #2749 comment quoting session `msg_4f7f536d` to PM `deep-wolf-155`. Asks dispositioned: -1. **(c-refined) substrate-shape — RATIFIED**. `EmissionTarget` sum +1. **(c-refined) substrate-shape — RATIFIED**. `WorkflowRuntime` sum (🟡 YELLOW per §7.3.1) in gunbc namespace; `project_github_actions: - (CIWorkflowDag, EmissionTarget) → Workflow` in gunbc namespace; + (CIWorkflowDag, WorkflowRuntime) → Workflow` in gunbc namespace; `extdeps.github.actions.Workflow` unmodified; `Workflow` values strictly derived as projection outputs. 2. **PR #2746 disposition — AMEND**. still-heron-763 strikes §3.1 diff --git a/docs/design-section-1-8-acceptance-aggregator-pattern.md b/docs/design-section-1-8-acceptance-aggregator-pattern.md index 1a2303b85b3..917b630dbdd 100644 --- a/docs/design-section-1-8-acceptance-aggregator-pattern.md +++ b/docs/design-section-1-8-acceptance-aggregator-pattern.md @@ -90,13 +90,13 @@ There is no stored `Status` value for §1.9 views. The Derived Status is always Proposed §1.9 entry (post-(c-refined) canvas ratification per PR #2749; constituent gate-IDs reflect PM cascade renames including `ci_yml_deleted` → `ci_yml_hand_authority_dissolved`): ``` -| V1 | t_ci_wad_full_r3_close | T-CI-WAD | depends_on: #56, #, #, #, #, #, # | | Brian-elevated to FULL R3-close scope 2026-05-12 per gunbc#846. Constituents: ci_workflow_modeled_as_dag + ci_yml_hand_authority_dissolved + emission_target_open_enum_landed + test_cost_dimension_landed + slow_test_exemptions_dissolved + project_github_actions_landed + ci_uses_affected_set_selection. | +| V1 | t_ci_wad_full_r3_close | T-CI-WAD | depends_on: #56, #, #, #, #, #, # | | Brian-elevated to FULL R3-close scope 2026-05-12 per gunbc#846. Constituents: ci_workflow_modeled_as_dag + ci_yml_hand_authority_dissolved + workflow_runtime_open_enum_landed + test_cost_dimension_landed + slow_test_exemptions_dissolved + project_github_actions_landed + ci_uses_affected_set_selection. | ``` The 6 NEW constituent rows MUST be added to **§1.8** (not §1.9) with their canonical families per the existing §1.8 schema: - `ci_yml_hand_authority_dissolved` → state-check (PM cascade per PR #2744 commit 19a1d8dfc — file presence orthogonal to hand-authority dissolution) -- `emission_target_open_enum_landed` → substrate-shape (gunbc-namespace per (c-refined)) +- `workflow_runtime_open_enum_landed` → substrate-shape (gunbc-namespace per (c-refined)) - `test_cost_dimension_landed` → substrate-shape (verify row number against existing cost-dim work) - `slow_test_exemptions_dissolved` → state-check - `project_github_actions_landed` → substrate-shape (new; projection function in gunbc namespace per (c-refined))