Open d-e-s-o opened 1 year ago
Thousands separators don't seem too hard to add. See #705. As for the padding/alignment, here is the relevant logic in libtest
, for reference:
I believe given the current state of the reporter logic that this requires a bit of a restructuring of internals to be feasible. Right now the reporter does not know of test names ahead of time from what I gather. If you have any suggestions, please let me know.
Thanks for adding the
bencher
output format. I have bothlibtest
based benchmarks as well as criterion based ones, and I use this output format to have better comparability.However, I noticed that the formats don't quite match up (which I will say kind of runs against the stated purpose: "it may be useful to support external tools that can parse this output"). Specifically, this is output generated by
libtest
benchmarks:Compare that to criterion based ones:
(some of those
criterion
benchmarks are in different groups; I haven't checked whether that has any impact)There are two main differences that I can make out:
Would it make sense and be possible to align the two more closely?