dan sinclair
dan sinclair
In [7.4.4 For Statement](https://www.w3.org/TR/WGSL/#for-statement) we say: - If update_part is non-empty, it becomes a continuing statement at the end of the loop body. and - The identifier of a declaration...
The `for` statement section [7.4.4 For Statement](https://www.w3.org/TR/WGSL/#for-statement) says: > If condition is non-empty, it is checked at the beginning of the loop body and if unsatisfied then a § 7.4.6...
Can I have an `@align(1)` or does the align value have to be a multiple of the natural alignment for the type? (So, i32 has to be 4, 8, 16,...
In [4.4.6 Address Space](https://www.w3.org/TR/WGSL/#address-space) the table lists the `uniform` restriction as `Constructable Host-shareable types`. This requires diffing the two lists to determine what that actually means. In general, it's the...
In the Types section (https://www.w3.org/TR/WGSL/#vector-types) and (https://www.w3.org/TR/WGSL/#matrix-types) we list some key use cases. This is only done for the vector and matrix types and seems out of place given nothing...
In the `TYPE INFERENCE FOR LITERALS` we say: ``` // When the context does not force a concrete type, an integer literal is // interpreted as an i32 value: ```...
When reading the types section, you get to `Plain Types` and then want to see the non-plain types but the next section of the spec is `Memory` and then we...
We previously unreserved type names in other languages. The following list still exist as reserved, which, if any, should be removed? * 'RWBuffer' * 'RWByteAddressBuffer' * 'RWStructuredBuffer' * 'RWTexture1D' *...
In [6.16 Constant Identifier Expression](https://www.w3.org/TR/WGSL/#constant-identifier-expr) the table refers to both `let` and `override` identifiers but the table caption specifically says `let-declared` only. The first row is talking about `override` but...
In 12.5 we state that an implementation doesn't have to consider infinity and can return any valid value. We then have several places in the spec which say things can...