Iron out possible failures associated with randomized spec order #2483
Description
We recently enabled random spec ordering, which leads to better robustness in our specs longterm:
I believe they are indeed revealing random failures based on ordering. I believe we should:
- Seek to confirm that failures are indeed due to random ordering
- Possibly temporarily switch back to non-random ordering while we get the issues resolved
- Resolve the issues related to random ordering
Thoughts?