Description

Our WinUI Device tests are prone to crashing because of a couple upstream WinUI bugs, so we need to structure the tests to be less prone to failing and able to recover. For example, if you open more than 255 windows the app will crash so there's somewhat of a ticking time bomb with our tests as they grow.

Possible resolutions

• Update to 1.5 Bump the windowsappsdk group with 3 updates #20460 (this is less important than getting our pipeline green but if we think updating to 1.5 will make it easier to get our pipeline green)
  • The new version might have some fixed to NavigationView that makes it less crash heavy
  • We were able to update to 1.4 successfully but one of our tests broke Bump the windowsappsdk group with 3 updates #18606. WinAppSDK has indicated they aren't going to fix this in 1.4 or 1.5 so we need to disable those tests. We could just start by bumping to 1.4 for now. That way we have a WinAPPSDK bump in the next SR.
• Setup the device tests so that tests are ran individually or in very small batches. Can we just run a single test per run? Can we just retry a single test?
• We currently run all the categories separately but in one pipeline. Should we run each category in a separate pipeline? Then we can just retry single categories.