* Excommunicate `PoolAllocator`
* Switch to manual deallocation
* Move `ResourceMemory` away from the `suballocator` module
* Remove `SuballocatorError::BlockSizeExceeded`
* Fix examples
* Fix atom size
* Address safety TODOs
* Nice English you got there bro
* ValidationError-ify commands, add some validation to `UnsafeCommandBufferBuilder`
* Use `SubpassBeginInfo` for the triangle example
* ValidationError-ify queries
* Switch to a more expressive way to specify memory usage
* Fix tests
* Fix examples
* Forgot to adjust `SubbufferAllocatorCreateInfo::location_preference`s
* Oopsie
* Redo all of it
* Oopsie
* Add `RequiresAllOf`, automatically enable required extensions and features
* Add more missing backticks and backslashes to error messages
* Use updated VUIDs that require `acceleration_structure`
* ValidationError-ify `Surface` and `Swapchain`
* Update vulkano/src/lib.rs
Co-authored-by: marc0246 <40955683+marc0246@users.noreply.github.com>
* Update vulkano/src/lib.rs
Co-authored-by: marc0246 <40955683+marc0246@users.noreply.github.com>
* Don't draw if the window has zero size
---------
Co-authored-by: marc0246 <40955683+marc0246@users.noreply.github.com>
* Reorganise render pass objects, add support for some extensions
* Handle stencil initial/final layout in the macro
* This was promoted to 1.2, not 1.1
* Doctests again...
* Missing docs
* Fix bad validation
* Convert Framebuffer to ValidationError while we're at it
* FramebufferCreateFlags
* Make render pass creation more like Vulkan again
* TODO for depth_resolve_mode and stencil_resolve_mode in the macro
* Depth/stencil resolve modes
* Missed validation
* Forgot some testing code...
* msaa-renderpass docs
* Doc fixes
* Add missing backslashes
* Updated #2208
This commit solves the problem of freezing when going fullscreen for MacOS.
It was caused because winit may behave unexpectedly when using
run_return. Winit crate generally discourages the use of that
(see: <https://docs.rs/winit/latest/winit/platform/run_return/trait.EventLoopExtRunReturn.html#tymethod.run_return>).
NOTE. This commit only solves the problem for `interactive_fractal`
binary, but not the `multi_window_game_of_life` binary.
* Some display servers (like X.org) are not redrawing for every frame.
* Fixed the same bug for example binary.
* Remove ControlFlow::Poll (it is the default anyway) in interactive_fractal
Co-authored-by: marc0246 <40955683+marc0246@users.noreply.github.com>
* Remove ControlFlow::Poll (it is the default anyway) in multi_window_game_of_life
Co-authored-by: marc0246 <40955683+marc0246@users.noreply.github.com>
---------
Co-authored-by: marc0246 <40955683+marc0246@users.noreply.github.com>
* Merge `SyncCommandBuffer` into `*AutoCommandBuffer`
* Commands take containers directly instead of iterators
* Use allocator as the type parameter for built command buffers
* Don't make `UnsafeCommandBuffer` keep objects alive
* Re-add impl for Arc
* Fixes
* Doc fix
* Replace `ImageAccess::descriptor_layouts()`
* Add `WriteDescriptorSet` constructors for image views with layout
* Replace `descriptor_layouts` for real
* Small changes
* More improvements
* Rewrite shader and specialization handling in pipelines
* Make the shader loading in examples a bit cleaner
* Forgot some
* Fix incorrect color blend states in examples
* Nicer fix
* Use mem::discriminant
Co-authored-by: marc0246 <40955683+marc0246@users.noreply.github.com>
---------
Co-authored-by: marc0246 <40955683+marc0246@users.noreply.github.com>
* Move `DeviceAlignment`
* Generate alignment-properties with the type `DeviceAlignment`
* Small fix
* Another small fix
* Switch to using `is_aligned` where possible
* Oopsie
* Add `BufferContentsLayout`
* Rework `BufferContents`
* Add `BufferContents` derive macro
* Relax restrictions of shader code generation
* Fix examples
* Add additional invariant to `Subbuffer`
* Shorten paths a bit
* Remove a bit of bloat
* Add `Sized` constraint to element traits
* Fix an oopsie
* Add `Aligned`
* Add support for deriving `BufferContents` for sized types
* Fix alignment and padding issues
* Fix docs and add examples for `BufferContents`
* Adjust shader macro
* Add tests
* Adjust `Vertex` example
* Remove bytemuck re-export
* Update examples
* Workaround bytemuck's array elements that are `AnyBitPattern` limitation
* Add more alignments
* Fix an earlier oopsie
* Rework vulkano-shaders
* Fix examples
* Fix some rogue tabs in examples
* Add `AsRef` and `AsMut` implementation for `Padded`
* Remove useless code duplication
* Make the `BufferContents` derive macro the same for all types
* Add example docs for `Padded`
* Work around trivial bounds
* More example docs
* Minor consistency adjustment
* Add `serde` to the list of cargo features
* Make clippy happy again
* Implement VertexDefinition for VertexBufferInfo, change Vertex signature to return VertexInfo and implement VertexInput-trait helper for Vertex to construct VertexBufferInfo objects from Vertex types. Updated teapot example to show usage.
* Reimplement BuffersDefinition leveraging VertexBufferInfo
* Reduce the amount of types required, remove VertexInfo, expose VertexBufferInfo methods on Vertex and use std HashMap transparently instead.
* Fix teapot example
* Fix outdated docs and use full path to Vertex trait to avoid issues
* Fix formatting of imports
* Remove clone and directly instantiate VertexInputBindingDescription.
* Rename VertexBufferInfo to VertexBufferDescription and being update of examples to use it
* Run cargo clippy fix to cleanup imports in examples
* Remove obsolete comment
* cargo fmt
* Refactor Vertex trait to not rely on ShaderInterfaceEntryType::to_format and instead rely on Format provided by VertexMember trait.
* Add test for impl_vertex macro, remove tuple implementations as they do not implement Pod, minor cleanups to impl_vertex macro.
* #[derive(Vertex)] proc-macro implementation with support for format and name attributes. Tests are implemented for both attributes and inferral matching impl_vertex macro
* Rename num_elements into num_locations to make purpose clear, add helper function to calculate num_components and check them properly in BufferDefinition's VertexDefinition implementation.
* Rename num_locations back to num_elements to make distinction to locations clear. Updated VertexDefinition implementation for BuffersDefinition to support double precision formats exceeding a single location.
* Add additional validation for vertex attributes with formats exceeding their location.
* Collect unnecessary, using iterator in loop to avoid unnecessary allocations.
* Use field type directly and avoid any form of unsafe blocks.
* Match shader scalar type directly in GraphicsPipelineBuilder
* Rename impl_vertex test to fit macro name
* Add VertexMember implementatinos for nalgebra and cgmath (incl matrices).
* Add missing copyright headers to new files in proc macro crate
* Document derive vertex with field-attribute options on the Vertex trait
* Add example for vertex derive approach.
* Do not publish internal macros crate as it is re-exported by vulkano itself
* Deprecate impl_vertex and VertexMember and update documentation for Vertex accordingly
* Make format field-level attribute mandatory for derive vertex
* Update all examples to derive Vertex trait instead of impl_vertex macro
* Fix doctests by adding missing imports and re-exporting crate self as vulkano to workaround limitations of distinguishing doctests in proc-macros
* Use actual integer bitflags for Vulkan bitflag types
* Fix doc tests
* Fix Fence and Semaphore on Win32 and Fuchsia
* Add macro for bitflags that have a corresponding enum
* Fix bug
* impl present_wait feature
* handle timeout case
* add missing doc
* document the return type of wait_for_present
* impl check for VUID-04999
* don't enforce gt if present_id is zero
* impl PresentInfoExt
* Remove NonZeroU64 from wait_for_present in favor of an error
* fix gl-interop example
* fix docs and fmt
* PresentInfoExt -> PresentInfo, Added swapchain + index to info
* fix doc test, didn't save file...
* fix gl-interop
* Remove Whitespace
* opt-in portability
* Only enable portability flag if extension is present
Automatically enable khr_portability_subset on Device; Remove required_extensions from PhysicalDevice & Device
* enable InstanceCreateInfo.enumerate_portability on all examples
* add additional documentation for enumerate_portability
* Added ImageUsage::transient_input_attachment()
* Address TODOs in examples
* Added actual example to command_buffer module documentation.
* Fixed error in command_buffer example
* Check max_framebuffer_height/width/layers in AttachmentImage constructors.
* Add detailed explanation for shader! macro to triangle examples
* Use only the selected range of buffer and image views in synchronization
* Improve render pass commands, more validation
Co-authored-by: Austin Johnson <me@austinj.work>
* Allow `SyncCommandBufferBuilder` commands to specify the access range
* Allow `SyncCommandBufferBuilder` commands to specify the access range
* Better handling of image initialization
* Small fix
* Rewrite copy commands
* Better handling of maintenance1
* Small leftovers
* Remove conflict_key
