Closed
Description
A mere cargo test
neither compiles nor runs the benchmarks. In order to keep them from bitrotting, they must be built and run in CI. But release builds take much more time and RAM than debug builds. In a CI situation, the benchmark results aren't important; only the fact that they can run without errors. So it would be very helpful if there were a cargo bench --debug
option
The closest that can be done right now is to do cargo build --benches && target/debug/<my_benchmark>-*
. That usage is problematic because
- The CI script must be updated everytime a new benchmark program is added, and
- That invocation will run old, stale binaries, too.
Instead, it would be great if cargo had a single cargo bench --debug
command that would build the benchmarks in debug mode and then run them.