Materia 0.4.1.0

Added

• Shared GLTF/GLB asset cache: GLTFLoader now uses a shared same-URL GLTFAssetCache by default, including in-flight load deduplication so repeated loader instances can reuse one fetch/decode pass.
• Safe GLTF instancing API: GLTFAsset.instantiate() creates independent scene graphs with separate object transforms and hierarchy while intentionally sharing geometry, material, texture, and animation resources for renderer reuse.
• Minimal GLB binary parsing: GLTFLoader can now parse GLB 2.0 containers with embedded binary buffer chunks, enabling direct repeated .glb cache coverage.
• WebGL readback regression guard: Added JS browser coverage that asserts steady WebGLRenderer.render() frames do not call readPixels.

Changed

• Asset resolver cache scoping: AssetResolver now exposes an optional cacheKeyScope; platform default resolvers share a stable scope, while custom resolvers remain isolated unless callers opt into a shared scope.
• JS browser test opt-in: Root jsBrowserTest can now be enabled with -PenableJsBrowserTests=true instead of being permanently disabled in the Gradle configuration.
• Packaging version metadata: Linux, macOS, and Windows packaging scripts now target 0.4.1.0.

Fixed

• Repeated GLTF/GLB loads: Loading the same model URL through separate GLTFLoader instances no longer repeats top-level model fetch/decode work when the cache scope matches.
• Cached scene mutation safety: Cached GLTF source assets are not returned directly to callers, preventing repeated instances from sharing mutable transform state.

Tests

• Added common loader tests for same-URL cache reuse, concurrent in-flight deduplication, transient failure eviction, cloned instance independence, shared render resources, and GLB embedded-buffer loading.
• Added JS compile/browser coverage for WebGL steady-frame readback behavior.