Closed abrown closed 1 year ago
I can not reproduce the hex-simd failure with CI, can you? Seeing that we don't have access to those log files, in situations like this it may be nice to print the diff between the expected output and the actual output.
After talking to @alexcrichton about hex-simd
, I pushed 3c061ec to make the benchmark more robust to changes in the WebAssembly binary. I think we can safely ignore the rebuild failure: it's spidermonkey
, which can't rebuild itself (#231).
I think we can safely ignore the rebuild failure: it's
spidermonkey
, which can't rebuild itself (#231).
Can we make CI green by ignoring spidermonkey rebuilds or something here? I don't want to get used to CI always being red because "CI isn't passing because X reason" and then that hides Y regression that we don't notice from landing.
Emscripten- and wasi-sdk-built projects already use a common
Dockerfile
for rebuilding the benchmarks. This change extends that to the Rust-built benchmarks by symlink-ing toDockerfile.rust
. The idea is to reduce duplication and make it easier to change build configuration in one place. For example, this change also bumps the used version of Rust to 1.70, which causes the emitted WebAssembly to slightly change.