DavidSpickett wrote: > Do you have any suggestions on how we should handle the "expected" flakiness > because of how we test the debugger? Do you think this is something we should > try to solve as part of the lldb testing framework?
Super basic tip, in case you didn't notice already, Linaro's experience is that running LLDB testing on a shared machine is a bad idea. For AArch64 we run on a dedicated machine and that has worked well. It's not very large, it's just free of distractions. For Arm we are on a shared server with much more resource than we need for LLDB, but we can't always get it because of other workers. So we set a parallelism limit way below the actual core count so that lit doesn't over estimate what it can run. This leaves us with the actually flakey tests that ideally we'd find proper fixes for as Jonas said. https://github.com/llvm/llvm-project/pull/129817 _______________________________________________ lldb-commits mailing list lldb-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
