Enum Layout SSOT & Take-Project Hardening

4 sections

Overview

Centralize ALL enum layout knowledge into a single source of truth in ori_repr, migrate every codegen consumer to use that SSOT instead of hardcoded layout assumptions, harden the AIMS take-project ownership model against lineage gaps and iterator-only scope limitations, and replace weak source-text-only niche tests with emitter-driven IR verification — resolving 10 confirmed dual-source TPR findings from the repr-opt §07 architectural review and making the enum representation architecture sound for all current AND future layout optimizations (niche filling, payload compression, §08-§12 work).

○

Planned

4 sections

Section 1 Not Started

EnumLayoutInfo API

Create a canonical EnumLayoutInfo query surface in ori_repr that answers ALL enum layout questions — tag encoding, payload sizing, field offsets, ABI size — eliminating 8+ scattered packing computations

0/25 tasks

Section 2 Not Started

TagAccess Full Migration + Derive Fix

Migrate ALL 88 codegen consumer sites to use TagAccess + EnumLayoutInfo, eliminating every hardcoded enum field index — fixing the active tagged-pointer miscompile in derive enum bodies

0/35 tasks

Section 3 Not Started

Take-Project Hardening

Enforce the in_class ⊆ var_to_lineage invariant via lineage reconciliation and generalize is_take_project from iterator-only tag whitelist to MachineRepr-driven ownership classification

0/28 tasks

Section 4 Not Started

Verification & Testing

Replace source-text niche-helper tests with emitter-driven IR verification, add codegen gate audit ensuring no TagAccess bypass, and run full integration verification

0/35 tasks