Fix/skip integration tests #10959
Fix/skip integration tests #10959
Conversation
There was a problem hiding this comment.
Classifications seem to have changed 🤷♂️
Interesting! I would expect "class name" to come from Roslyn semantic ranges and "RazorTagHelperElement" to come from Razor. I know you said everything looked OK, but it worries me that "class name" never shows up. Does that mean that this is actually a regression and tag helper elements are no longer colorized using the "User Types - Classes" Fonts and Colors setting? Instead of updating base lines in integration tests with unexpected results, is there a deeper regression that should be investigated?
Yes, and since opening the PR I believe this is likely a FUSE-related regression. The reason I think it's worth changing the baseline for these specific tests, which I definitely should have mentioned in the description, is because the classification in question here is largely unrelated to the tests. The tests are specifically validating project load when a Razor buffer is created early in the startup lifecycle, and the fact that they both failed caused me to start investigating that path, which is long and complicated. Turns out, it's just that these are the only two integration tests that wait for the "class name" classification 🤦♂️ By changing the tests to wait for a Razor defined classification we should be more immune to changes, either due to Roslyn or to the generated code, in these specific tests. There are existing tests that validate classifications are as we expect, and I plan on adding coverage to them for FUSE. I had thought that was already done somewhere but my hunch now is that it's not.
Ironically the opposite. Tag helpers still appear purple (for me and whatever theme I have), and this change validates that moving forward. The "class name" classification would have applied to the text after the |
|
Okay, I have found our test that covers semantic tokens in |
|
Pushed up test changes so our semantic tokens tests now cover |
|
Skipped a couple more tests that already have FUSE bugs logged. |
Oof... I didn't think test machines got TNs. I wonder if there's a setting to disallow that... |
|
eh, I don't think its a bad thing necessarily. These test failures essentially found https://devdiv.visualstudio.com/DevDiv/_workitems/edit/2268945 before CTI did (not that we noticed or acted in time 😛) Having control over feature flags would be good though, to get FUSE and non-FUSE runs, and that would have made it immediately obvious where FUSE was causing issues, but we've talked about that many many times. |
|
Integration tests: ✅ |
Cherry pick of the commit from #10959 that fixes code actions in FUSE, **and fixing the generate event handler code action which is unrelated to FUSE**, in case we decide we want to take it for 17.12 P3 Fixes https://devdiv.visualstudio.com/DevDiv/_workitems/edit/2268945
We have 9 failing integration tests, seemingly because the test machines are picking up FUSE. This PR skips some (with links to bugs) or works around the issues.
Fixes https://devdiv.visualstudio.com/DevDiv/_workitems/edit/2268945