Open klahnakoski opened 4 years ago
@klahnakoski should we move this issue to https://github.com/armenzg/smart-scheduling ?
The idea would be to have two views: one comparing the schedulers on all pushes, one comparing the schedulers on backstop pushes only.
The SETA-scheduled labels may impact the determination of regression; showing SETA as more efficient than the other schedulers. If we limit our analysis to pushes with all tasks scheduled, we may get more accurate numbers for those other schedulers.