[thehobbit85]

This PR is fully superseded by upstream commit 2aae9f8c4e054d2178743a9fbec29075089eddd2, included in fixed main 635e8008d67053214ddd302a0196f5ec971b63ea. It adds the same instruction binding in ue_region.py and the same size-register assertion in patch_nso.py. Both complete files are byte-identical between this PR head and current main. There is no remaining source change to carry forward; closure is recommended. The PR remains open and its branch has not been rebased or rewritten.

Fresh verification of those upstream sources: the pinned Python lint ratchet passes with 392 baseline findings, zero new findings and zero stale entries. The focused UE-region/lint/docs tests are also included in the fresh packing validation recorded on #376; historical tests below retain their original tested revisions.

Historical PR content and validation on earlier revisions

Main still has the two Unreal engine-region lint findings: the operand helper captures the loop instruction, and the observation gate shifts a possibly-optional size-register value. Bind the instruction and assert the detector's size-register invariant, without changing the lint baseline or valid rewrite behavior.

Maintenance pass (2026-10-31, 01:10 UTC): rebased onto fixed main f11018a1f312, producing 39752d582c7f. No conflicts; range-diff confirms the patch is unchanged. #376 also includes the same fix as an author-added prerequisite commit; this standalone PR keeps the correction available independently.

Fresh validation on 39752d582c7f:

• Native arm64 Python 3.14: python3 -m unittest dm.tools.tests.test_ue_region dm.tools.tests.test_lint_python dm.tools.tests.test_docs — 28 passed.
• Pinned ruff/pyright ratchet against fixed main: 392 baseline findings, 0 new, 0 stale; baseline unchanged.
• git diff --check passes. No game-module execution or iPhone validation was performed.

Exact fixed-main f11018a1f312 → 39752d582c7f Z-A shader/interface comparison: PASS on chain d1043b780a74. All 28,603 shaders build in both revisions, with 0 changed MSL, new failures or interface issues. The completed shared cache and current reflection hashing were used; this comparison started no shader translation. Z-A has no smoke route, so this is not gameplay or rewritten-module execution coverage.
The local Z-A build is not Unreal-based; the focused synthetic UE-region tests remain the detector-specific verification.

Fresh exact-base shader/interface report

regress.py check (full): PASS

Baseline f11018a1f3125ec3046c30faae6147956a1a7c60 f11018a, candidate 39752d582c7f24787c9e401e3e2f3be19f79a856 39752d5; Apple M2 Max, macOS 26.6, Xcode 27.0; 85.7 s

game	status	shaders (base → cand)	newly failing	newly building	changed MSL	interface (new / known)	boot	fps p50	frame p99 ms	peak MB	prepare s	new RED
LegendsZA 0100F43008C44000	PASS	28603/28603 → 28603/28603	0	0	0	0 / 0	–	–	–	–	–	–

LegendsZA 0100F43008C44000: PASS

• time: chain 85.4 s

Original investigation, wider tool tests and historical regression results

What and why

lint-python fails on main at 2006d5ed with two findings introduced by the Unreal engine-region detector. The initial audit found the same inherited failures on #221, #376 and #684; their feature changes do not cause either finding.

Bind the current Capstone instruction in the operand helper, and assert the detector supplied a size register before encoding the observation gate. This makes the existing invariants explicit and leaves the lint baseline unchanged. Valid rewrite behavior is unchanged.

This is a standalone fix against main so the unrelated PRs do not have to own the underlying correction. Equivalent fixes reached #221 and #684 concurrently; #376 now includes this fix as a separate prerequisite commit. This standalone PR makes the correction available to main and other contributors.

Checked

• Reproduced the two failures on main with the pinned tools: ruff 0.17.2 / pyright 1.1.414; 394 findings against 392 baseline findings.
• After the fix: python3 dm/tools/lint_python.py --against origin/main passes with 392 baseline findings, zero new findings and zero stale entries.
• All 68 dm/tools/tests/test_*.py files passed (698 tests reported: 682 passed, 16 skipped), including the seven synthetic Unreal engine-region cases and documentation/lint-ratchet tests.
• git diff --check passed; independent review found no issues.
• scripts/regress.py impact origin/main..HEAD reports broad rewriter reach. Only Pokémon Legends: Z-A (0100F43008C44000, build v393216) is discoverable locally, so the regression check covered that build. Its profile has no smoke route: the report below checks shaders, not gameplay or rewritten-module execution. No five-game or iPhone validation is claimed. The synthetic tests exercise the detector's Unreal-specific instruction patterns.
• Device: Apple M2 Max, macOS 26.6, Xcode 27.0.

regress.py check (full): PASS

Baseline 2006d5ed8690d79ef4865a60917eb938cdb4c8dc 2006d5e, candidate worktree 2006d5e (+ uncommitted changes); Apple M2 Max, macOS 26.6, Xcode 27.0; 28.4 s

game	status	shaders (base → cand)	newly failing	newly building	changed MSL	interface (new / known)	boot	fps p50	frame p99 ms	peak MB	prepare s	new RED
LegendsZA 0100F43008C44000	PASS	28603/28603 → 28603/28603	0	0	0	0 / 0	–	–	–	–	–	–

LegendsZA 0100F43008C44000: PASS

• time: chain 28.2 s

Docs

The existing engine-region mechanism remains documented in memory.md. This is an invariant/lint correction, with no new runtime mechanism or shader-chain change. No game files, keys or extracted data are included.

[seedds]

Thanks for isolating these shared lint failures. Closing this PR as superseded by 2aae9f8.

Main contains the instruction binding in ue_region.py and the size-register invariant assertion in patch_nso.py. The PR's latest verification reports both files identical to main, with no remaining source change to merge.