unify(ww3d2): Merge dx8renderer, dx8rendererdebugger

[stephanmeesters]

Summary of additions/fixes added to Generals (AI generated):

Source	Change
dx8rendererdebugger	The DX8 renderer debugger was moved into Core/ so it can be shared.
dx8renderer	Texture-stage handling was aligned with MeshMatDescClass.
dx8renderer	Delayed procedural material-pass support was added for rigid meshes.
dx8renderer	Delayed translucent passes are now rendered during the final flush.
dx8renderer	Logging and renderer-debugger support were added.
dx8renderer	A skin-container allocation leak was fixed.
dx8renderer	Decal rendering was guarded against an empty visible-decal list.
dx8renderer	Sorting criteria were updated for meshes with explicit sort levels.
dx8renderer	Unregistering a mesh now also removes its gap filler.

[greptile-apps]

Greptile Summary

This PR moves dx8rendererdebugger to Core/ so it can be shared, aligns texture-stage loops to MeshMatDescClass::MAX_TEX_STAGES, adds a delayed procedural material-pass pipeline for rigid meshes (so translucent-only passes render after all base geometry), fixes a skin-container double-allocation, guards decal rendering against an empty list, updates sort-level criteria, and removes gap fillers on mesh unregister.

• P1 — render state leak: In DX8TextureCategoryClass::Render(), D3DRS_NORMALIZENORMALS is set TRUE before the new debugger guard but reset FALSE inside it (lines 1862-1863 vs. 1929-1930). Any scaled mesh that the renderer debugger disables leaves the state dirty, corrupting lighting on all subsequent meshes.

Confidence Score: 4/5

Safe to merge after fixing the D3DRS_NORMALIZENORMALS state leak introduced by the new debugger guard.

One P1 finding: the NORMALIZENORMALS reset is now inside the new debugger if-block while the set remains outside, creating a dirty render state when the debug tool is active on a scaled mesh. All other changes (delayed pass infrastructure, decal guard, gap-filler cleanup, sort-level fix, texture stage alignment, file move) are correct and well-structured.

Generals/Code/Libraries/Source/WWVegas/WW3D2/dx8renderer.cpp — specifically the NORMALIZENORMALS state bracketing around the new debugger guard in DX8TextureCategoryClass::Render().

Important Files Changed

Filename	Overview
Generals/Code/Libraries/Source/WWVegas/WW3D2/dx8renderer.cpp	Major feature additions — delayed procedural pass system, debugger guard, decal null guard, gap-filler cleanup, sort-level fix. Contains a P1 render state leak: D3DRS_NORMALIZENORMALS is set TRUE before the new debugger if-block but reset FALSE inside it, leaving dirty state when a scaled mesh is disabled by the debugger.
Generals/Code/Libraries/Source/WWVegas/WW3D2/dx8renderer.h	Adds AnyDelayedPassesToRender flag, pure-virtual Add_Delayed_Visible_Material_Pass/Render_Delayed_Procedural_Material_Passes to the base class, implements them in Rigid and Skin containers, and switches texture array size to MeshMatDescClass::MAX_TEX_STAGES. Clean changes.
Core/Libraries/Source/WWVegas/WW3D2/dx8rendererdebugger.h	Renamed/moved from GeneralsMD/ to Core/ with no content changes; uses #pragma once correctly.
Core/Libraries/Source/WWVegas/WW3D2/CMakeLists.txt	Enables dx8rendererdebugger.cpp/.h in the Core WW3D2 build by removing the comment-out markers.
GeneralsMD/Code/Libraries/Source/WWVegas/WW3D2/CMakeLists.txt	Comments out dx8rendererdebugger in the GeneralsMD local build now that it lives in Core/.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[DX8MeshRendererClass::Flush] --> B[Render rigid FVF containers]
    B --> C[DX8FVFCategoryContainer::Render]
    C --> D[Render visible texture categories]
    D --> E{Debugger enabled AND mesh disabled?}
    E -- No --> F[Render mesh: sorting or immediate]
    E -- Yes --> G[Skip mesh render]
    F --> H[Render_Procedural_Material_Passes]
    G --> H
    H --> I[Render skin FVF containers]
    I --> J[Render_Decal_Meshes guarded: early return if empty]
    J --> K[Render_FVF_Category_Container_List_Delayed_Passes NEW]
    K --> L[DX8RigidFVFCategoryContainer::Render_Delayed_Procedural_Material_Passes]
    L --> M[Clear vertex/index buffers]

Loading

Prompt To Fix All With AI

This is a comment left during a code review.
Path: Generals/Code/Libraries/Source/WWVegas/WW3D2/dx8renderer.cpp
Line: 1869-1936

Comment:
**D3DRS_NORMALIZENORMALS state leak when debugger skips a mesh**

`D3DRS_NORMALIZENORMALS` is set to `TRUE` at line 1862-1863 (before the new debugger guard), but the corresponding reset to `FALSE` lives at lines 1929-1930 — inside the new `if (!DX8RendererDebugger::Is_Enabled() || !mesh->Is_Disabled_By_Debugger())` block. When the debugger is active and disables a mesh whose `ObjectScale != 1.0f`, the `FALSE` path is never reached, leaving the render state dirty for all subsequent meshes in the loop (causing their normals to be re-normalized, which visibly alters lighting/shading).

Move the `NORMALIZENORMALS` reset outside the debugger guard so it mirrors the unconditional set:

```cpp
if (!DX8RendererDebugger::Is_Enabled() || !mesh->Is_Disabled_By_Debugger()) {
    // ... existing render code ...
}
//--------------------------------------------------------------------
if (mesh->Get_ObjectScale() != 1.0f)
    DX8Wrapper::Set_DX8_Render_State(D3DRS_NORMALIZENORMALS, FALSE);
//--------------------------------------------------------------------
```

How can I resolve this? If you propose a fix, please make it concise.

_{Reviews (2): Last reviewed commit: "Remove empty line" | Re-trigger Greptile}

[xezon]

Merging and moving need to be at least 2 separate commits that go into main branch. Meaning it is not legal to squash and merge this change.

[stephanmeesters]

Merging and moving need to be at least 2 separate commits that go into main branch. Meaning it is not legal to squash and merge this change.

OK that did not happen with shdlib.h in #2499 unfortunately.

Do you want to split this into two PR's or shall I turn the it into two commits for merge by rebase (after review)?

[xezon]

After looking more closely I see that the debugger files existed in ZH only. So moving them in one change is fine.

[xezon]

Looks ok.

[xezon]

greptile has raised 2 potential issues. Perhaps address them in a follow up change if necessary.

[stephanmeesters]

greptile has raised 2 potential issues. Perhaps address them in a follow up change if necessary.

See #2610