You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Tests and builds have some fragility and and do not seem to repeatedly, reliably run with deterministic outcomes.
It looks like there are a couple of issues converging to cause this problem. The first is race conditions in tests and benchmarks, particularly around parallel benchmark runs that attempt to build artifacts that are locked by other runs. The second is that bindable properties created with the source generator were using ConfigureAwait(false) which combined with async void was exacerbating the above mentioned race conditions.
Note that I have investigated only the apparent failures and don't have a good understanding of the full context, so will appreciate some input and guidance here.
Expected Behavior
Tests should deterministically pass or fail based on expected conditions, not ephemeral environmental conditions.
Activity