- Updated to NET 8 and Aardvark.Base 5.3
- Use struct representation for partial active patterns
- Removed Effect.toLayered
- Initial prerelease
- [GLSL] Fixed issue with variable names containing @
- [Raytracing] Fixed issue with callable data name indexing
- [Raytracing] Added duplication checks in effect builders
- [Debugger] Fixed issue with compiler options not being recognized due to double hyphens
- Added Effect.toLayered with custom semantic
- Use std430 layout for all storage buffers
- [GLSL] Fixed issue with image format names containing underscores
- Removed warning on field set for storage buffers
- Added caching for raytracing shaders and effects
- Made hashes of compute shaders more consistent
- Added check for type redefinitions in module compiler
- [Debugger] Fixed issue with single unit input shader functions
- [Debugger] Disabled symbol and other miscellaneous file generation on build
- [Debugger] Enable file watchers on demand due to inotify limit on Linux
- [Debugger] Fixed backslash handling in paths
- [Serializer] Include more type information in hashes to avoid conflicts when using shader debugger
- Reworked shader debugger (see https://github.com/aardvark-platform/aardvark.docs/wiki/FShade-ShaderDebugger)
- Reduced the number of OnlyInShaderCode exceptions thrown by the optimizer
- Fixed constant folding of bitwise operations on enums
- Added validation for uniform aliases
- Serializer includes uniform semantic and type for hashing
- Added sampler filter reduction mode
- [RTX] Added builder operation overloads with untupled arguments
- [RTX] Added RayHitKind enum
- [RTX] Ray flags and hit kind inputs use enums now
- Turned Module into a class to hide laziness of entries
- Added support for enums with arbitrary underlying type
- Added config for reversing tesselation winding order
- Preprocessor: merge aot2 + bugfixes
- Lazy modules entries
- Effect serialization bugfixes: array deserialization, uniform scopes
- Fixed issue with return type for integer vector dot product
- Include sampler texture name in hash
- Added support for float32-based vertex types
- Added sampler filter reduction mode
- [RTX] Added builder operation overloads with untupled arguments
- [RTX] Added RayHitKind enum
- [RTX] Ray flags and hit kind inputs use enums now
- Turned Module into a class to hide laziness of entries
- Added support for enums with arbitrary underlying type
- Added config for reversing tesselation winding order
- Preprocessor: merge aot2 + bugfixes
- Lazy modules entries
- Effect serialization bugfixes: array deserialization, uniform scopes
- Fixed issue with return type for integer vector dot product
- Include sampler texture name in hash
- Added support for float32-based vertex types
- Initial prerelease version for 5.4
- fshadeaot using resources instead of literal strings
- merged v52 (containing improved AOT)
- Added intrinsics for dynamically accessing matrix rows and columns (matrix.Row() / matrix.Column())
- Added intrinsics for dynamically accessing matrix elements (matrix.[x, y])
- Added missing intrinsics for special floating point checks (NaN, +-infinity)
- Added missing intrinsics for vector swizzles in Vec module
- Added and improved intrinsics for vector and matrix relations (AllEqual, AllDifferent, ...)
- Added intrinsics for DistanceSquared, Distance1, DistanceMin, DistanceMax, Norm1, Norm2, NormMin, NormMax
- Fixed return type for dot product of integer vectors
- Fixed issue with deserialization of raytracing interfaces
- Added simple hash based id to RaytracingEffect
- Added matrix transform intrinsics
- Removed redundant dimension parameter in CNewVector
- Fixed issue with reflected functions using sampler types as parameters
- Removed unused / unnecessary image and sampler types
- Remove unused image formats
- Remove MipMapLevels property for multisampled samplers
- Allow multiple InterpolationMode values
- Fixed various image and sampler methods and properties
- Added Samples property for multisampled images and samplers
- Added SampleLevelOffset
- Implemented implicit flat interpolation for integral types
- Fixed and added vector and matrix intrinsics
- Fixed various intrinsics with regard to duplicated expressions
- Added exp2 intrinsic
- Added step, linearstep intrinsics
- Implemented basic output type conversion (e.g. from V4d to V2d)
- Added Debug.Printfn (Vulkan via validation layers only)
- Implemented full support for unsigned integer types (images, samplers, output types)
- Added support for unsigned integer types (output, samplers, images, ...)
- Removed interpolation modes for built-in fragement inputs
- Removed MipMapLevels for multisampled samplers
- Implemented Debug.Printf
- Fixed regular expression for constant swizzles
- Improved error message for incompatible output types
- Initial prerelease version for 5.3
- Fixed issue with constant swizzles
- disabled warnings again
- printing warnings when Expr.TryGetReflectedDefinition fails
- try/with for Expr.TryGetReflectedDefinition (seems to raise exceptions in some cases)
- Fixed issue with topological sort and raytracing data
- Removed warnings in raytracing shaders
- Updated to Aardvark.Base 5.2
- Aardvark.Build version
- aardpack
- aardpack