Implement WGSLLanguageExtension
#6814
Conversation
sagudev
changed the title
ImplementedLanguageExtension as WGSLLanguageExtensionWGSLLanguageExtension
|
cc @ErichDonGubler, who wrote #6437. |
sagudev
force-pushed
the
wgsl-language-features
branch
from
8787ee3 to
e9ab502
Compare
|
rebased and squashed into two commits (one for naga changes, one for wgpu changes). |
sagudev
force-pushed
the
wgsl-language-features
branch
from
e9ab502 to
b89fe77
Compare
| const ALL: [ImplementedLanguageExtension; 0] = []; | ||
|
|
||
| /// Returns slice of all variants of [`ImplementedLanguageExtension`] | ||
| pub const fn all() -> &'static [Self] { | ||
| &Self::ALL | ||
| } |
There was a problem hiding this comment.
nitpick: It would be more future-proof to implement this in terms of strum::{EnumIter,IntoEnumIterator}, and return an impl Iterator<Item = ImplementedLanguageExtension instead.more maintenance-proof.
There was a problem hiding this comment.
Oh didn't know we already have strum as dep, will fix this.
There was a problem hiding this comment.
Actually it's only dev-dep and it's not present in FF, are we still okay to add it?
There was a problem hiding this comment.
I did it anyway in 0bc74e2 using VariantArray. I will revert it if we do not want strum as dep.
| bitflags::bitflags! { | ||
| /// WGSL language extensions. | ||
| /// | ||
| /// WGSL spec.: <https://www.w3.org/TR/WGSL/#language-extensions-sec> | ||
| #[derive(Debug, Clone, PartialEq, PartialOrd, Ord, Eq, Hash)] | ||
| pub struct WGSLLanguageFeatures: u32 { | ||
| /// <https://www.w3.org/TR/WGSL/#language_extension-readonly_and_readwrite_storage_textures> | ||
| const ReadOnlyAndReadWriteStorageTextures = 1 << 0; | ||
| /// <https://www.w3.org/TR/WGSL/#language_extension-packed_4x8_integer_dot_product> | ||
| const Packed4x8IntegerDotProduct = 1 << 1; | ||
| /// <https://www.w3.org/TR/WGSL/#language_extension-unrestricted_pointer_parameters> | ||
| const UnrestrictedPointerParameters = 1 << 2; | ||
| /// <https://www.w3.org/TR/WGSL/#language_extension-pointer_composite_access> | ||
| const PointerCompositeAccess = 1 << 3; | ||
| } | ||
| } | ||
|
|
There was a problem hiding this comment.
question: Is there a reason this shouldn't be in wgpu-types? This would also have the benefit of not needing to add the bitflags dep. to wgpu.
There was a problem hiding this comment.
Not really, except that it is only used in wgpu.
| "readonly_and_readwrite_storage_textures" => { | ||
| Some(crate::WGSLLanguageFeatures::ReadOnlyAndReadWriteStorageTextures) | ||
| } | ||
| "packed_4x8_integer_dot_product" => { | ||
| Some(crate::WGSLLanguageFeatures::Packed4x8IntegerDotProduct) | ||
| } | ||
| "unrestricted_pointer_parameters" => { | ||
| Some(crate::WGSLLanguageFeatures::UnrestrictedPointerParameters) | ||
| } | ||
| "pointer_composite_access" => { | ||
| Some(crate::WGSLLanguageFeatures::PointerCompositeAccess) | ||
| } |
There was a problem hiding this comment.
issue: Matching on identifiers here duplicates logic that we already have LanguageExtension::from_ident.
suggestion: We're also going to need to map from ImplementedLanguageExtension to WGSLLanguageFeatures anyway when using the wgpu_core backend, so let's match LanguageExtension::from_ident(wlf.as_str()) (probably using a separate method on WGSLLanguageFeatures or an implementation of Into) instead.
There was a problem hiding this comment.
The problem is that here is webgpu impl, that does not necessary have naga available.
But we do use LanguageExtension::from_ident for parsing in core backend.
ErichDonGubler
added
area: api
Signed-off-by: sagudev <[email protected]>
Signed-off-by: sagudev <[email protected]>
Signed-off-by: sagudev <[email protected]>
Signed-off-by: sagudev <[email protected]>
Signed-off-by: sagudev <[email protected]>
sagudev
force-pushed
the
wgsl-language-features
branch
from
b89fe77 to
ce36775
Compare
Signed-off-by: sagudev <[email protected]>
Connections
#6350
Description
This is to allow browsers to implement https://www.w3.org/TR/webgpu/#gpuwgsllanguagefeatures, currently we do not implement any such extension (but I am working on pointer_composite_access, hence this PR).
Testing
There is none.
Checklist
cargo fmt.taplo format.cargo clippy. If applicable, add:--target wasm32-unknown-unknown--target wasm32-unknown-emscriptencargo xtask testto run tests.CHANGELOG.md. See simple instructions inside file.