In https://github.com/psf/pyperf/commit/c9604435bc46273f3b6e51101ea4091d57a2020b, which fixed a bug in scaling the mem_max_rss value on Darwin, I inadvertently changed the measurement from using RUSAGE_SELF to RUSAGE_CHILDREN. This has resulted in the measurements being much coarser and incorrect on Darwin. This reverts to the old correct behavior.
In https://github.com/psf/pyperf/commit/c9604435bc46273f3b6e51101ea4091d57a2020b, which fixed a bug in scaling the
mem_max_rss
value on Darwin, I inadvertently changed the measurement from usingRUSAGE_SELF
toRUSAGE_CHILDREN
. This has resulted in the measurements being much coarser and incorrect on Darwin. This reverts to the old correct behavior.