mirror of
https://github.com/intel/llvm.git
synced 2026-01-20 01:58:44 +08:00
1e3af94253
It seems the fail fast just doesn't strike the right balance. It wastes too many resources, especially if a build is killed because the machine it was running on got preempted. Instead, we should simply not run any future jobs if a failure has occured, while letting the already running jobs finish.