Generalize the cuda-bias test cases by replacing hardcoded "cuda" literal with the DEVICE variable

[EikanWang]

As the title.

[EikanWang]

I checked all failures. It was due to a newly added line of code to skip a few test cases on XPU - from helion._testing import skipIfXPU. The newly added line of code changes the line number information.

Should I update the expected file directly? @yf225

[oulgen]

I checked all failures. It was due to a newly added line of code to skip a few test cases on XPU - from helion._testing import skipIfXPU. The newly added line of code changes the line number information.

Should I update the expected file directly? @yf225

Run tests with EXPECTTEST_ACCEPT=1 to automatically update expected files

[oulgen]

you need to rebase one more time

[oulgen]

@EikanWang you're still having some line number disparity, please rebase to main and run

EXPECTTEST_ACCEPT=1 pytest test/test_type_propagation.py

[EikanWang]

Thanks. I will find a CUDA machine and do it.

[EikanWang]

@jansel, @oulgen, may I know the merge process for an approved PR? Can I merge it directly if the CI signal is green? Or is there any other process that I need to follow?

[oulgen]

@jansel, @oulgen, may I know the merge process for an approved PR? Can I merge it directly if the CI signal is green? Or is there any other process that I need to follow?

i merged for you. Did you have merge permissions?

[oulgen]

@EikanWang is there an XPU machine we can use for the CI?

[EikanWang]

@oulgen , we are setting up the XPU machines and will enable XPU CI as long as these machines are ready.

[jansel]

@jansel, @oulgen, may I know the merge process for an approved PR? Can I merge it directly if the CI signal is green? Or is there any other process that I need to follow?

You can merge if CI is green and you have an approval.