Open psfinaki opened 1 week ago
There might not be one common reason for all of these tests other that they don't work on core CLR. It is probably not worth the time investment to look at ~150 tests individually and try to rewrite each of them for core CLR. Most/all of them were migrated from fsharpQA.
I don't think there is a common reason yeah. But it might be worth at least visually going through that and analyzing what's especially concerning (like - what really should work on Core CLR).
I personally don't think it's worth spending time on it. Not now for sure. Majority of the tests were migrated from desktop suite (fsharpqa or cambridge suites), and likely rely on desktop compiler, bcl and runtime behaviour. To run them everywhere they'll likely need to be duplicated, corrected and verified.
FYI I am not talking about those ones, the migrated ones are component tests. Also good to look at but yeah can wait.
I meant those compiler service tests, and it's only about a dozen of them - mostly in ProjectAnalysisTests
and newly discovered AssemblyContentProviderTests
.
We have already a few dozens of compiler service tests marked as
FactForDesktop
.While skimming through them, I didn't find a good reason for that in most cases, they should be revisited and either fixed or given a proper explanation on why they are disabled.