erik/acdream
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 1

fix(pipeline): MP1a - sink delegate restores immediate side-stage enqueue (exception-path faithfulness)

Browse source 
Coordinator-directed follow-up. The buffer-and-drain seam diverged from
the original on the exception path: pre-MP1a, CollectEmittersFromScript
enqueued particle-preload meshes DIRECTLY into _stagedMeshData
mid-Prepare, so preloads staged before a later throw in the same
Prepare* call (reachable via PrepareEnvCellMeshData side-staging during
its StaticObjects loop, then PrepareCellStructMeshData throwing on a
malformed-dat texture decode) were already safely enqueued. The drain
version only flushed after a successful return — on throw, entries
stranded on the shared extractor until an unrelated successful call
flushed them, and were silently dropped on dispose.

Fix: MeshExtractor takes an Action<ObjectMeshData>? sideStagedSink
constructor parameter; the two CollectEmittersFromScript sites become
_sideStagedSink?.Invoke(meshData) — the original code shape (immediate
hand-off) at those exact lines. ObjectMeshManager wires the sink to
_stagedMeshData.Enqueue, restoring the original immediate-enqueue
semantics including on mid-Prepare throw. _sideStaged buffer,
DrainSideStaged(), and the ProcessQueueAsync drain loop are deleted.
The MP1b bake tool passes its own collector.

Inventory doc updated: MP1a note now records the sink seam and the
Content-owned upload enums, so its no-behavior-change claim is accurate.

dotnet build green; full test suite 4059 passed / 0 failed / 4 skipped.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
Erik 2026-07-05 20:39:42 +02:00
parent 5722681bf5
commit 932f904e00
3 changed files with 39 additions and 37 deletions
Download patch file Download diff file Expand all files Collapse all files

18
docs/architecture/worldbuilder-inventory.md
View file

@ -53,8 +53,26 @@ behavior change, no divergence-register row.
- `src/AcDream.Content/TextureKey.cs` — the atlas dedup key, lifted out of
the GL-owning `TextureAtlasManager` (which stays in App and now
references the lifted struct).
- `src/AcDream.Content/UploadFormats.cs` — Content-owned
`UploadPixelFormat`/`UploadPixelType` enums carried by
`MeshBatchData`/`TextureBatchData` instead of
`Silk.NET.OpenGL.PixelFormat`/`PixelType` (Content must stay
Silk.NET-free — the bake tool must not ship GL binaries). Underlying
values are the GL ABI constants, numerically identical to the Silk.NET
members; App casts at its single upload boundary (the `AddTexture` call
in `UploadGfxObjMeshData`) via a lifted nullable enum conversion —
value- and null-preserving.
- `src/AcDream.Content/IDatReaderWriter.cs`, `EdgeLineBuilder.cs` — GL-free
dependencies of the extractor, moved (namespace-only) alongside it.
- **Side-stage sink seam:** `CollectEmittersFromScript` pre-loads particle
GfxObj meshes mid-extraction and, pre-MP1a, enqueued them directly onto
`ObjectMeshManager._stagedMeshData`. The extractor now takes an
`Action<ObjectMeshData>? sideStagedSink` constructor parameter; App wires
it to `_stagedMeshData.Enqueue`, preserving the original
immediate-enqueue semantics exactly — including on a mid-`Prepare*`
throw (preloads staged before a malformed-dat texture-decode exception
survive, as they always did). The MP1b bake tool passes its own
collector.
- **Stays in `src/AcDream.App/Rendering/Wb/`:** `ObjectMeshManager` (now a
thin wrapper owning the staged-queue/worker-pool/Dispose-quiesce lifecycle
and all GL upload — constructs one `MeshExtractor` and delegates every

13
src/AcDream.App/Rendering/Wb/ObjectMeshManager.cs
View file

@ -167,7 +167,12 @@ namespace AcDream.App.Rendering.Wb {
_graphicsDevice = graphicsDevice;
_dats = dats;
_logger = logger;
_extractor = new MeshExtractor(_dats, _logger);
// Side-stage sink: particle-preload meshes staged mid-extraction go
// straight onto the staged-upload queue, exactly as the pre-MP1a code
// did — immediate enqueue, surviving a later throw in the same
// Prepare* call. ConcurrentQueue.Enqueue is thread-safe for the
// extractor's up-to-4 concurrent decode workers.
_extractor = new MeshExtractor(_dats, _logger, data => _stagedMeshData.Enqueue(data));
_useModernRendering = _graphicsDevice.HasOpenGL43 && _graphicsDevice.HasBindless;
if (_useModernRendering) {
GlobalBuffer = new GlobalMeshBuffer(_graphicsDevice.GL);
@ -475,12 +480,6 @@ namespace AcDream.App.Rendering.Wb {
// If it's a direct setup or gfxobj, make sure background loads don't abort half-way
data = _extractor.PrepareMeshData(id, isSetup, CancellationToken.None);
}
// MP1a: drain any mesh data the extractor staged as a side effect
// (particle emitter GfxObj preloads inside CollectEmittersFromScript)
// and enqueue it the same way the top-level result is enqueued below.
foreach (var sideStaged in _extractor.DrainSideStaged()) {
_stagedMeshData.Enqueue(sideStaged);
}
if (data != null) {
lock (_cpuMeshCache) {
if (_cpuMeshCache.Count >= _maxCpuCacheSize) {

45
src/AcDream.Content/MeshExtractor.cs
View file

@ -39,38 +39,23 @@ public sealed class MeshExtractor {
private readonly ThreadLocal<BcDecoder> _bcDecoder = new(() => new BcDecoder());
/// <summary>
/// MP1a mechanical seam: retail's particle-preload side effect (see
/// <see cref="CollectEmittersFromScript"/>) used to enqueue directly onto
/// ObjectMeshManager's runtime staged-upload queue (<c>_stagedMeshData</c>).
/// That queue is a GL-upload-lifecycle concern and stays in App per the
/// plan's binding rules, so the extractor collects the same side-effect
/// output here instead; ObjectMeshManager drains it via
/// <see cref="DrainSideStaged"/> immediately after each Prepare* call and
/// enqueues the results itself. No behavior change — same objects reach
/// the same queue, just via an explicit hand-off instead of a shared field.
/// ConcurrentQueue because one MeshExtractor is shared by up to
/// MaxParallelLoads (4) decode workers — the original enqueued to the
/// thread-safe _stagedMeshData, so the hand-off buffer must be
/// thread-safe too.
/// MP1a mechanical seam: receives particle-preload meshes staged
/// mid-extraction (see <see cref="CollectEmittersFromScript"/>). The App
/// wires this to its staged-upload queue, restoring the original
/// immediate-enqueue semantics — entries must survive a subsequent throw
/// in the same Prepare* call (retail's code enqueued directly onto
/// ObjectMeshManager's <c>_stagedMeshData</c> mid-Prepare, so preloads
/// staged before a malformed-dat texture-decode throw were already safe).
/// The MP1b bake tool passes its own collector. The sink must be
/// thread-safe: one MeshExtractor is shared by up to MaxParallelLoads (4)
/// decode workers.
/// </summary>
private readonly ConcurrentQueue<ObjectMeshData> _sideStaged = new();
private readonly Action<ObjectMeshData>? _sideStagedSink;
public MeshExtractor(IDatReaderWriter dats, ILogger logger) {
public MeshExtractor(IDatReaderWriter dats, ILogger logger, Action<ObjectMeshData>? sideStagedSink = null) {
_dats = dats;
_logger = logger;
}
/// <summary>
/// Drains and returns any mesh data staged as a side effect of the most
/// recent Prepare* call (particle emitter GfxObj preloads). See
/// <see cref="_sideStaged"/>.
/// </summary>
public List<ObjectMeshData> DrainSideStaged() {
var result = new List<ObjectMeshData>();
while (_sideStaged.TryDequeue(out var item)) {
result.Add(item);
}
return result;
_sideStagedSink = sideStagedSink;
}
/// <summary>
@ -176,13 +161,13 @@ public sealed class MeshExtractor {
if (emitter.HwGfxObjId.DataId != 0) {
var meshData = PrepareMeshData(emitter.HwGfxObjId.DataId, false, ct);
if (meshData != null) {
_sideStaged.Enqueue(meshData);
_sideStagedSink?.Invoke(meshData);
}
}
if (emitter.GfxObjId.DataId != 0 && emitter.GfxObjId.DataId != emitter.HwGfxObjId.DataId) {
var meshData = PrepareMeshData(emitter.GfxObjId.DataId, false, ct);
if (meshData != null) {
_sideStaged.Enqueue(meshData);
_sideStagedSink?.Invoke(meshData);
}
}
}