Conversation
|
we, uutils, use codspeed could you please copy what is done with coreutils ? |
|
CodSpeed does not count some syscalls. How about your suggestion? |
|
@sylvestre : I am not sure what you mean by "copy". Could you please elaborate. I just migrated to codspeed criterion drop-in replacement. But would you like me to drop Criterion and use the divan benchmark instead? At this early stage this would not be too hard. |
|
yes, please use divan |
|
I removed Criterion and migrated the tests to Divan. |
|
I found the performance.md for coreutils and will rework this now to match that logic. |
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #185 +/- ##
===========================
===========================
☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
| bencher | ||
| // .with_inputs(|| prepare::cmp_params_identical_testfiles(lines)) | ||
| .with_inputs(|| params.clone()) | ||
| .bench_refs(|params| black_box(cmp::cmp(¶ms).unwrap())); |
There was a problem hiding this comment.
sorry, i approved it too quickly. you should not call it as blackbox
the stack isn't currently interesting:
it should call the rust function directly
There was a problem hiding this comment.
Ok, will remove blackbox
3 participants
During development it is often helpful to be able to quickly benchmark the recent changes.
Criterion offers the ability to bench single functions.
This PR introduces criterion and has benches for the parsers (negligible) and cmp compare.
Has also a support function to generate large files to compare for bench-marking.
Update: Changed to Codspeed with Divan.