Skip to content

Commit 1a0afe3

Browse files
committed
docs(portability): reconciliation note — scan ran pre-seam; cure already on this branch
Persist the reconciliation, not the raw scan artifact (raw scan would be a stale competing plan). Scan's 5 blockers == the seam swap already applied here: 2547c83 (Option B, repoint 5 tsconfigs to local base, drop @repo/typescript-config) + 97bd271 (strictness fixes); devDep dropped from 4 of 5 (matches scan's own count); isolated-clone build proven 34/34 with @repo absent from lockfile/store. Live actions: (1) merge extraction/oss-seam + collapse the double clone (platform clone canonical), (2) verify validate:publish/changeset in standalone CI, (3) optional @loopengine no-hyphen alias cleanup; then re-run the scan on the merged ref to produce the evidenced-READY gate artifact.
1 parent 97bd271 commit 1a0afe3

1 file changed

Lines changed: 49 additions & 0 deletions

File tree

Lines changed: 49 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,49 @@
1+
# Loop Engine build-portability scan — reconciliation note
2+
3+
> **Status:** reconciliation note (OSS track). **Do not commit the raw scan artifact.** It scanned a
4+
> **pre-seam ref**; its blockers are already cured on `extraction/oss-seam`. Committing the raw scan
5+
> would create a competing, stale plan — the same duplicate-plan failure mode the GATEWAY work hit.
6+
> What is worth persisting is *this* note: what the scan found, why it is already addressed, and the
7+
> few **live** actions that remain.
8+
9+
## TL;DR
10+
11+
The portability scan ran against the OSS repo **before the seam swap**. Every blocker it raised is the
12+
exact `@repo/*`-shared-config dependency the seam branch already removed. The cure is committed; the
13+
isolated-clone proof already passed. The scan is therefore **stale, not wrong** — it is evidence *for*
14+
the seam branch, not a new work plan.
15+
16+
## What the scan found vs. what is already done
17+
18+
| Scan finding (pre-seam) | Reality on `extraction/oss-seam` |
19+
| --- | --- |
20+
| 5 packages depend on shared `@repo/*` build config (`@repo/typescript-config`) | **Cured.** `2547c83` — Option B: repoint the 5 `tsconfig`s to a **local base** + parity flags + DOM lib; drop `@repo/typescript-config`. |
21+
| Strictness/parity gaps once off the shared base | **Cured.** `97bd271` — strictness fixes for the 5 packages under the unified local base. |
22+
| Scan's own count: the dev-dependency was present on **4 of 5**, not all 5 | **Matches.** the devDep was dropped from the **4** packages that actually carried it — the scan's 4-not-5 finding is consistent with the cure, not a discrepancy. |
23+
| Will it build standalone with `@repo/*` absent? | **Proven.** isolated-clone build passed **34/34**; `@repo` absent from both the lockfile and the pnpm store in the clean clone. |
24+
25+
**Net:** the scan's five blockers ≡ the swap already applied on `extraction/oss-seam` (`2547c83`
26+
config-only + `97bd271` strictness), with the isolated-clone proof already green.
27+
28+
## Live actions that remain (the only forward work)
29+
30+
1. **Land/merge `extraction/oss-seam` and collapse the double clone.** The OSS code currently exists in
31+
two clones; the **platform clone is canonical** (already a checklist precondition). Merge the seam
32+
branch, then collapse to the single canonical clone so there is no second source of truth.
33+
2. **Verify `validate:publish` / changeset runs in standalone CI** on the merged ref (the scan checks
34+
the package graph, not the release pipeline — confirm publish validation is green outside the
35+
monorepo).
36+
3. **(Optional) Clean up the legacy `@loopengine` no-hyphen aliases** — cosmetic/name-hygiene, not a
37+
build blocker.
38+
39+
## Then: re-run for the gate
40+
41+
Once `extraction/oss-seam` is merged and the double clone is collapsed, **re-run the portability scan
42+
on the merged ref** to produce the **evidenced-READY** artifact for the OSS build-portability gate.
43+
That re-run — against post-seam state — is the artifact to keep; this note is superseded by it.
44+
45+
## Why a note and not the artifact
46+
47+
The raw scan describes a state that no longer exists. Persisted as-is it becomes a competing plan that
48+
re-asserts already-cured blockers (the duplicate-plan failure mode). The durable record is: **scan ran
49+
pre-seam → cure committed on the seam branch → re-run post-merge for the gate.**

0 commit comments

Comments
 (0)