Closed dabreegster closed 3 months ago
Changes look good but I've just noticed some tests are breaking. I wonder how that's been introduced or if it's some flakiness? Although we don't see much flakiness on GH these days
Hmm, I also saw some timeout flakiness running locally. I'll look into this a bit.
While looking into flakiness, there were many legitimate test failures in savefiles
. Rethinking things, the previous fix is confusing in v1 mode, because it hides the problem of not selecting an intervention type, even though the user can control it. So I've changed the fix to only work in pipeline mode, where this is no radio button for that field.
Running individual tests locally is fine with low parallelism. Let's see what a second round of GH actions looks like. They only use 2 workers, according to logs
… external file