You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
#6938 (use hashbrown in more crates) causes cargo nextest run in the naga subdirectory to fail:
cargo nextest run
Compiling naga v24.0.0 (/home/jimb/wgpu/naga)
error[E0277]: the trait bound `hashbrown::set::HashSet<spirv::Capability, BuildHasherDefault<rustc_hash::FxHasher>>: Deserialize<'_>` is not satisfied
--> naga/tests/snapshots.rs:45:19
|
45 | capabilities: naga::FastHashSet<spirv::Capability>,
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ the trait `Deserialize<'_>` is not implemented for `hashbrown::set::HashSet<spirv::Capability, BuildHasherDefault<rustc_hash::FxHasher>>`
|
I think the issue here is that serde itself implements Deserialize for std::hash::HashSet, but not hashbrown::HashSet.
Apparently CI doesn't run Naga tests without features.
The text was updated successfully, but these errors were encountered:
Add `hashbrown` with the `"serde"` feature as a development dependency
for Naga. Regardless of whether Naga's `deserialize` feature is
enabled, the snapshot tests need to deserialize parameters saved in
files as RON text.
It would also suffice to use `std::collections::HashSet` in the
snapshot tests, although we would need to build a `naga::FashHashSet`
from the std `HashSet` at one point. Adding the dev-dependency seems
slightly simpler.
Fixesgfx-rs#7038.
My one suggestion is that you can do nextest on a single crate from the project root directory with a command like this:
cargo nextest run -p naga
and yes, this IS also broken by #6938. I would personally favor both updating CI to test these & update README to clarify this. I would be happy to contribute these updates, if needed.
#6938 (use hashbrown in more crates) causes
cargo nextest run
in thenaga
subdirectory to fail:I think the issue here is that serde itself implements
Deserialize
forstd::hash::HashSet
, but nothashbrown::HashSet
.Apparently CI doesn't run Naga tests without features.
The text was updated successfully, but these errors were encountered: