rad experiment show cc506f895d7390f61e60e34565e0fbefef4c3528
by ee@did:key:z6MkvsiybCuk1WDZhh2aXDx7NTiZejKgWZygMNsUuXDMNvVQ · Apr 20 11:36 2026
Try workspace `resolver = "3"` to see if newer feature-resolution rules reduce clean cargo check work further.
Measurements
Metric
BaselineThe benchmark measurement of the unmodified code
CandidateThe code with the proposed optimization applied
Candidate/Parent ΔThe performance change from baseline to candidate
cargo_check_s (s)primary
Reduce the wall-clock time of a clean `cargo check --all` for this workspace on the current macOS host.
The benchmark represents edit-compile iteration speed for the Rust workspace, with emphasis on changes in `wgpu-native` and `wgpu-remote` source/layout that reduce compile work without breaking the build.
5.300 s (n=1)
5.798 s (n=1)
+9.39%REGRESSED
target_kbsecondary
80948.000 (n=1)
80948.000 (n=1)
0.00%
Annotations
hypothesis
Cargo resolver 3 may cut additional feature or host-dependency unification beyond resolver 2 on modern Cargo.
learned
For this workspace and toolchain, resolver 3 did not improve the clean check workload over resolver 2.
next_action_hint
Stay on resolver 2 and investigate dependency or crate-structure changes with clearer mechanisms.
rollback_reason
The benchmark regressed versus the resolver-2 baseline with no artifact-size improvement.
Base CommitThe starting commit before the optimization was applied