Eric Wilhelm wrote: > What's that calculation look like? From what I've seen, we currently > only have the harness start/end times.
Yeah, it's basically just doing a time() before the main runtests() and then again after it. > Maybe I'm unclear, but it sounds like something is going to break when > the archive is constructed of tests which were run in parallel and then > aggregated together. I'm not sure you'd call that a "break", it would just be a different metric when tests are run in parallel. Instead of being a sum of the durations of individual tests, it's now the total "real" time of the test run. > If we're going to calculate duration, it needs to be per-test. I agree that this is also a useful metric to have. But Smolder can only use the version of TAP that is currently being produced and there is no such information currently given. -- Michael Peters Developer Plus Three, LP
