[0.44 alpha.0] Prove the TSX-to-Part Program vertical compiler slice

[SisyphusZheng]

Part of #1155. Target: v0.44.0-alpha.0.

Objective

Prove the replacement architecture end to end before broad implementation.

Deliverables

• Compile one decorated OpenElement with text, property, event and conditional/list regions
• Use one generated Part Program for server HTML, fresh browser DOM and existing-DOM claim
• Integrate the spike through the real Deno/Vite transform path
• Measure generated code, allocations and activation against the 0.43 equivalent

Acceptance

• The spike contains no runtime VNode or BindingDescriptor discovery
• All three execution modes produce equivalent observable output
• A Signal updates only its subscribed Part/Region
• The result records GO/NO-GO evidence and unresolved compiler risks

Authority

• docs/current/VERSION_PLAN.md
• docs/adr/ADR-0143-0-44-compiled-element-model-reentry.md

[SisyphusZheng]

[ADR-0146 CLAIM]

Loop: a0-001
Candidate: 0.44.0-alpha.0
Base SHA: e31ab34eb8cea52fe07d6d138d0b0cfbba9d0f17
Acceptance slice: one decorated OpenElement transformed through the official Vite hook into one deterministic Part Program consumed by server serialization, fresh DOM and existing-DOM claim, with exact reactive sink updates and fail-closed unsupported syntax.
Dispatch: docs/evidence/v0.44.0-agent-loops/a0-001/dispatch.md
Owned paths: adapter internal compiler, the exact Vite plugin integration file, isolated compiler fixture/tests, element internal compiled-spike runtime/tests, and current loop implementer result.
Risk: critical

#1182 is closed at merge SHA e31ab34e; no other active loop owns these paths. The packet forbids public exports, private cross-package imports, workspace aliases, legacy fallback changes and control-plane edits.

[SisyphusZheng]

STATUS: PR_OPEN

Candidate SHA: bcc99b258c04190fcebd9a4c4db9f13d93e4754d
PR: #1184 targeting dev
Acceptance slice: a0-001 one-component-vite-part-program-vertical-spike

K3 implementer: PASS (session_5998df53-7c93-43ef-a13c-424763ba16cd)
Sol initial review: NO-GO-REPAIR
K3 repair-1: PASS
Sol repaired review: GO

Independent Sol evidence: packet gates all exit 0; adapter full 638 passed; element full 294 passed; repository deno task test exit 0; immutable base-SHA behavior RED probe exited 1 with the expected assertion failure.

Evidence: docs/evidence/v0.44.0-agent-loops/a0-001/

Next: wait for required PR checks, merge to dev if green, then close this issue at the merge SHA. This is not the alpha.0 version-closure GO.

[SisyphusZheng]

CORRECTION — candidate SHA

The exact candidate SHA for PR #1184 is bcc99b25df5a2ded515995b8fb0e3470bf87258e.

The earlier PR_OPEN comment contained an incorrect expanded SHA. The short SHA bcc99b25 and PR head were correct; this comment supersedes only that expanded SHA field.

[SisyphusZheng]

STATUS: CI_REPAIR_REQUIRED

Merged SHA: 4e1c322df76849846c8254946515e2575c758255
Failed candidate run: https://github.com/open-element/openelement/actions/runs/33071761401

AutoFlow exit 1. Exact failures: deno lint rejected the fixture button without an explicit type; deno task arch:check rejected three production as unknown as escapes. All other selected gates passed.

Repair packet: docs/evidence/v0.44.0-agent-loops/a0-001/repair-2.md
Issue remains open; this is an ordinary bounded K3 repair, not alpha.0 closure.

[SisyphusZheng]

STATUS: REPAIR_PR_OPEN

Candidate SHA: 18ac4c1057b6d69376d78c1d38e6aa9074576e50
PR: #1185 targeting dev
Slice: a0-001-repair-2

Implementer: PASS
Reviewer: GO
Independent verification: focused adapter 4/13, focused element 2/11, full adapter 638/71, full element 294/11; lint, architecture, typecheck, formatting all exit 0. Full AutoFlow retry passed 48/48 gates after one recorded npm registry network flake.

Next: wait for PR #1185 checks; merge only after required results are explicitly green, then close #1160 at the merge SHA.