Properly apply WGSL's automatic conversions to the arguments
to texture sampling functions.
Introduce helper function `Lowerer::expression_with_leaf_scalar`.
Although this can't affect behavior, use
`Lowerer::expression_for_abstract` for the `image` and `sampler`
arguments, simply because we want to move away from
`Lower::expression`'s automatic concretization, and move towards
having callers say explicitly what sort of conversions they need.
Although this can't affect behavior, use
`Lowerer::expression_with_leaf_scalar` for the `offset` argument, so
that the code spells out that this requires an `i32` value, rather
than depending on blind concretization giving it that.
Continue to use `Lowerer::expression` for `gather` and `array_index`,
since those happen to behave correctly with blind concretization, and
can be cleaned up later.
Fixes#7427.
Change `naga::front::wgsl::lower` to `use` `ir` and `proc`,
and then qualify identifiers with that instead of `crate`.
This is strictly fiddling with `use` directives and name
qualification, with the concomitant formatting changes; there should
be no change to the module's behavior or meaning.
Closes#7481.
This implementation roughly follows approach 2 outlined in #7481, i.e.,
it adds a polyfill for the signed and unsigned dot product of packed
vectors for each platform. It doesn't use the specialized instructions
that are available for this operation on SPIR-V (with capability
DotProductInput4x8BitPacked).
Have constant evaluator use `wrapping_abs` instead of `abs` when
applying `MathFunction::Abs` to AbstractInt values. WGSL says that
applying `abs` to the most negative AbstractInt value must return it
unchanged, which is what Rust's `i64::wrapping_abs` does.
Change `resolve_type` and `resolve_type_impl` to return
`TypeResolution`s. Add a new method `resolve_type_inner` that returns a
`TypeInner` (i.e. what `resolve_type` used to do).
