9aaf97e785
This reverts ceec3bc4. Two independent reasons, either sufficient.
The rendering regression. The slice deleted TextRenderGlStateScope, which
saved GL_MULTISAMPLE and GL_SAMPLE_ALPHA_TO_COVERAGE on entry, disabled them
for the text pass, and restored them on exit (TextRenderGlStateScope.cs:111-112
and 153-154 at the parent commit). Its replacement bakes that state into the
text pipeline but nothing restores it, and GlGpuPassEncoder.Dispose does not
either. Every world renderer is still raw GL at this point in the campaign, so
from the first UI frame onward the world drew with multisampling disabled.
The offline pixel gate caught it: 1,791 of 563,200 compared pixels differed,
0.318% against a 0.001 threshold. The commit message attributed this to
wall-clock-driven ambient animation shifting phase, and committed through the
failure. That explanation does not survive its own control: capturing twice at
the reverted-to commit differs by 19 pixels and twice at the slice's own commit
by 8, while base-versus-head differs by 1,791 - a 224x gap that no shared-noise
source explains. An amplified difference image settles it visually: the changed
pixels are the silhouette edges of every tree, building and rock, with terrain
interiors, water and the entire UI untouched. That is the signature of losing
edge antialiasing, not of animated sprites.
This is the exact failure mode two existing memory notes already warn about -
a mid-frame renderer must set every GL state it uses rather than inherit it,
and issue #52's lesson that a rendering migration must audit per-pass GL state
before declaring itself done.
The scope. The brief was three small leaf renderers plus additive frame-
lifecycle wiring, roughly ten files. The commit changed 334 files with 3,665
insertions and 3,845 deletions, including 323 public-to-internal visibility
conversions across the App assembly, 55 test files, two retired conformance
tests, and a self-described temporary escape hatch for bridging raw-GL viewport
textures. Even without the regression, that is not separable into the part
worth keeping and the part worth dropping.
Reverting rather than patching because the good work here - the RHI frame
lifecycle wiring and a genuine render-state-cache staleness fix - is small
enough to redo cleanly against a tightened spec, while untangling it from 300+
files of unrelated churn is not.
Post-revert: Release build clean, App suite back to 3,843 passed / 3 skipped,
offline pixel gate passing at 19 differing pixels.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
51 lines
3.1 KiB
C#
51 lines
3.1 KiB
C#
namespace AcDream.App.Streaming;
|
||
|
||
/// <summary>Result of the per-frame dungeon streaming-gate decision.</summary>
|
||
/// <param name="InsideDungeon">Passed to <see cref="StreamingController.Tick"/> — collapse
|
||
/// streaming to the single dungeon landblock.</param>
|
||
/// <param name="ObserverLandblockKey">When non-null, override the streaming observer to this
|
||
/// landblock key (the cell id's high 16 bits, 0xXXYY). Null leaves the caller's observer as-is.</param>
|
||
public readonly record struct DungeonGateResult(bool InsideDungeon, uint? ObserverLandblockKey);
|
||
|
||
/// <summary>
|
||
/// AP-36: the dungeon streaming gate (#133 FPS). When the player stands in a SEALED
|
||
/// EnvCell (an indoor cell that doesn't see outside), streaming collapses to the single
|
||
/// dungeon landblock — AC dungeons have no adjacent landblocks, so the normal 25×25
|
||
/// window would pull in ~129 unrelated ocean-grid dungeons. The trigger is the player's
|
||
/// CURRENT cell (<c>CellGraph.CurrCell</c>, set the moment the player is placed), and the
|
||
/// observer is pinned to that cell's OWN landblock (the cell id high 16 bits) because a
|
||
/// dungeon's EnvCells sit at arbitrary ocean-grid world coords with negative local offsets.
|
||
///
|
||
/// <para>Extracted from <c>GameWindow.OnUpdate</c> as a pure function so the
|
||
/// teleport-hold rule (below) is unit-testable without the GL/dat/network stack.</para>
|
||
/// </summary>
|
||
public static class DungeonStreamingGate
|
||
{
|
||
/// <summary>
|
||
/// Decide the streaming gate from the player's current cell.
|
||
/// </summary>
|
||
/// <param name="isTeleportHold">True while a teleport arrival is held (the controller is in
|
||
/// PortalSpace): the player is NOT yet placed, so <paramref name="currCellId"/> is the frozen
|
||
/// SOURCE cell, not where the player is going.</param>
|
||
/// <param name="currCellIsSealedDungeon"><c>CurrCell is EnvCell && !SeenOutside</c>.</param>
|
||
/// <param name="currCellId">The current cell id (0xXXYYNNNN).</param>
|
||
public static DungeonGateResult Compute(
|
||
bool isTeleportHold, bool currCellIsSealedDungeon, uint currCellId)
|
||
{
|
||
// #145/#138: during a teleport hold the player is NOT yet placed, so CurrCell is
|
||
// the frozen SOURCE cell — where the player IS, not where they're going. Streaming
|
||
// must follow the DESTINATION, which the PortalSpace observer pin already does, so
|
||
// the source-cell gate is suppressed. Otherwise a teleport OUT of a dungeon keeps
|
||
// streaming collapsed on the source dungeon (CurrCell still sealed) → the outdoor
|
||
// destination never hydrates → the TAS transit holds 600 frames → force-snap to
|
||
// ocean. A teleport INTO a dungeon is handled explicitly upstream by
|
||
// StreamingController.PreCollapseToDungeon (and the controller's _collapsed latch
|
||
// holds it through the hold), so suppressing the gate here doesn't regress it.
|
||
if (isTeleportHold)
|
||
return new DungeonGateResult(false, null);
|
||
|
||
if (currCellIsSealedDungeon)
|
||
return new DungeonGateResult(true, currCellId >> 16);
|
||
return new DungeonGateResult(false, null);
|
||
}
|
||
}
|