Update Cargo. #1406
Update Cargo. #1406
Conversation
Required for the bump in toml.
|
I'm assuming the CI failure is not related to this update? Are these tests particularly flaky on HFS? Trying locally, I can't repro on APFS, but happens frequently on HFS. We have a bunch of special cases in Cargo's testsuite to deal with this. I'm not sure if anyone has looked into it at all. |
|
That's actually very insightful! I don't have a macOS machine myself so I didn't delve deeper into that myself but yeah, we have some specific spurious macOS failures. @ehuss mind pointing me to the part of the Cargo's testsuite that's related to the HFS? |
|
I opened #1407 to show a temporary solution, which may give some hints. Generally you have to be careful with timestamps on HFS since they are such low resolution. In Cargo's testsuite, it detects if it is running on HFS with a very crude check. In the project builder, the mtime is set one second into the past here to ensure that building a project a second time will appear fresh. I tried that on the rls testsuite but it doesn't make a difference. Additionally we have some conditional sleeps for tests that need to rebuild the target multiple times. However, I don't think that's the case here, either. I see a few places in the RLS codebase that deal with time information. I tried investigating the one in rls-analysis but it doesn't seem related. I don't know enough about rls to investigate much further, but I suspect it is something similar to that line in rls-analysis. |
|
Thanks for the insight! 👍 I'll merge this if it's blocking you in any way and will come back to the HFS details later on, if you don't mind. |
Required for the bump in toml.