Open alrz opened 1 year ago
I think we typically use the Pairwise
attribute instead of Combinatorial
to keep the set of generated tests to a reasonable number. When combinatorial results in a test case explosion, randomly picking 30 could be a grossly inadequate coverage, whereas Pairwise (or 3-wise, etc.) provides a more provably confident coverage amount. Have you tried the Pairwise attribute?
Given a set of generated test cases, it's useful to be able to select a random subset for each run to reduce the CI time, while keeping the original combinatorial definition intact. e.g.
There's already a helper to do this
Random.GetItem
in dotnet 8p1 but I think it's easy to backport.