New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Remove CoreRT workaround #1643
Remove CoreRT workaround #1643
Conversation
This was fixed upstream.
@MichalStrehovsky we two few kinds of CoreRT integration tests:
Could you please update the hardcoded version here and here? This should make CI happy. Thank you! |
The CI failed with some new error (link):
|
Looks like a bad input to the compiler (assembly compiled against EntityFramework.dll, but EntityFramework.dll was not actually supplied as input). The compiler can sometimes survive this and generate a throwing method body wherever such thing is referenced, but sometimes it's hit in a spot that can't be recovered (like here - we end up needing it outside method compilation). JIT based runtimes will only hit trouble once this is hit in a code path at runtime, but AOT compilers don't have much of a choice. |
What surprises me is that we don't have an explicit dependency on I've no idea where it's coming from |
Usually it's a NuGet package with improperly declared dependencies (package forgetting to mention it depends on the EF package, but there's an assembly reference). |
Thank you! |
Now we are getting timeouts (BDN is configured with default 5 minutes timeout for the build). I've tried to repro it locally and on my machine, it takes a little bit more than 4 minutes but less than 5 and it works OK. I've checked our CI history and it looks like so far we neeeded +- 40 minutes to run all the tests and with this PR 1h is not enough. @MichalStrehovsky do you think that the new CoreRT compiler version could introduce a performance regression? |
Ah. We now use the same thing as .NET 5 - That would also explain the compiler crash. I assume we were trimming the problematic code previously because it was unused. |
@MichalStrehovsky that explains everything ;) I've pushed a fix, lets see if it helps |
@MichalStrehovsky We got to the point where Windows and macOS are passing, but Ubuntu fails with the following error:
FWIW The tests that are using hardcoded CoreRT version ( |
Thank you!
And sigh. I pushed out more verbose logging to find out what are we doing. |
@MichalStrehovsky have you changed anything in the product in the meantime? The build has passed this time oO |
We had changes but nothing targeted. Let's see what happens now. |
the CI failures are unrelated, I am merging. big thanks for all your help @MichalStrehovsky !! |
Thank you for maintaining the integration! |
This was fixed upstream.