It seems like we're formatting the test case arguments even on passing test runs, e.g. here is the flamegraph of a test run where the test passed, but we see 41% of runtime spent building up debug_reprs.
Does this make sense to you? do you see a way of modifying the test runner to only render the debug representation on failure?
It seems like we're formatting the test case arguments even on passing test runs, e.g. here is the flamegraph of a test run where the test passed, but we see 41% of runtime spent building up
debug_reprs
.Does this make sense to you? do you see a way of modifying the test runner to only render the debug representation on failure?