Measure 1,000- and 10,000-step traces in at least two browser engines.
Record the tested commit, browser, operating system, trace creation time, reverse-step latency, and peak memory. Split any correctness failure into its own issue.
These results will determine whether the next release can carry a defensible performance baseline.
Measure 1,000- and 10,000-step traces in at least two browser engines.
Record the tested commit, browser, operating system, trace creation time, reverse-step latency, and peak memory. Split any correctness failure into its own issue.
These results will determine whether the next release can carry a defensible performance baseline.