After http://trac.webkit.org/changeset/106211, we force gc() after each test run. However, we should also have a way to measure the time we spend in garbage collection and report it separately. e.g. if we some changeset make us spend 1000ms in gc instead of 100ms, then performance tests should capture that.
This is a nice-to-have feature but I wouldn't prioritize it at the moment.
(In reply to comment #1) > This is a nice-to-have feature but I wouldn't prioritize it at the moment. And it makes less sense for modern VMs which has generational/incremental GC. V8 has it. And JSC also has or is going to have one. What we need is, IMO, the profiler dump for each (or part of) test run.