rust-lang/rustc-perf

Expected end time calculation uses wrong time?

Mark-Simulacrum opened this issue · 1 comments

Saw this today, where the next run is expected to end +4 minutes from the currently ongoing run. My guess is that we're using the remaining time in the current run rather than the expected runtime of a run as the delta to add? Or perhaps ignoring that the current run is still ongoing?

image

Good catch! This was actually the combination of two separate bugs. #1802 should help.