Change printlns in the backend functions to say "generating" instead
of "writing" before calling the Naga backends. "Writing" suggests
writing output to a file; "generating" seems more accurate for calling
a backend.
Add a new type, `Input`, representing a particular Naga input file,
with methods to generate related paths - output files with a
particular extension, parameter files, and the input file itself.
Use this throughout `snapshots.rs` to generate paths.
Give `Input` utility methods for reading and writing files.
Rather than adding a `-glsl` suffix to the name, just rely on
`check_targets`' standard naming behavior, treating
`tests/in/variations.glsl` like any other snapshot input.
Remove the call to `Validator::validate` in `convert_spv`, since it
directly follows the call to `check_targets`, which also called
`Validator::validate`.
The only difference between the two is whether `Parameters::god_mode`
is respected, but this difference doesn't seem to have been
deliberate: at the time the call to `check_targets` was added to
`convert_spv` (5f21cf360, 2021-02-17), the two calls were exactly the
same.
Overriding the metal version in the workspace Cargo.toml using a patch
section does not work for projects building against the trunk version:
wgpu = { git = "https://github.com/gfx-rs/wgpu", branch = "trunk" }
Instead specify the metal dependencies only once in wgpu-hal.
Use `FastIndexSet`, rather than `FastHashSet`, for tracking global
declarations' dependencies, so that the order in which functions are
inserted into the `Module` is not dependent on the hash function.
* wgpu-core: Only produce StageError::InputNotConsumed on DX11/DX12
This error only exists due to an issue with naga's HLSL support:
https://github.com/gfx-rs/naga/issues/1945
The WGPU spec itself allows vertex shader outputs that are
not consumed by the fragment shader.
Until the issue is fixed, we can allow unconsumed outputs on
all platforms other than DX11/DX12.
* Add Features::SHADER_UNUSED_VERTEX_OUTPUT to allow disabling check
* Pick an unused feature id
Add a new Naga feature, `"compact"`, which adds a new function
`naga::compact::compact`, which removes unused expressions, types, and
constants from a `Module`.
The `front::wgsl::lowerer::RuntimeExpressionContext::local_table`
field does not need to be a mutable reference, as expressions never
introduce new local bindings.
Implements timer queries via write_timestamp on Metal for encoders (whenever timer queries are available) and passes (for Intel/AMD GPUs, where we should advertise TIMESTAMP_QUERY_INSIDE_PASSES now).
Due to some bugs in Metal this was a lot harder than expected. I believe the solution is close to optimal with the current restrictions in place. For details see code comments.
