fix: test to check that we can write via xrootd

[maxgalli]

This PR introduces a test to check that we can correctly write via XRootD (see #1252).
This test fails with current fsspec-xrootd, but it succeeds when running with the changes introduced in scikit-hep/fsspec-xrootd#76.
We can thus decide to either merge it now but skip the test, or wait until scikit-hep/fsspec-xrootd#76 is merged (and a new version released) and then merge it.

[maxgalli]

@ianna the fix was merged in fsspec-xrootd scikit-hep/fsspec-xrootd#76
what is the ideal procedure now? I guess:

• release a new version of fsspec-xrootd
• set a >= like dependency for fsspec-xroot here in uproot (set to the new release containing the bugfix)
• merge this PR with the new test

[ianna]

@ianna the fix was merged in fsspec-xrootd scikit-hep/fsspec-xrootd#76 what is the ideal procedure now? I guess:

• release a new version of fsspec-xrootd
• set a >= like dependency for fsspec-xroot here in uproot (set to the new release containing the bugfix)
• merge this PR with the new test

Thanks @maxgalli ! Yes, this is a reasonable suggestion. Let’s wait for the release and then merge it. Thanks.

[ianna]

@maxgalli - Thanks! It looks good to me. I wonder if we want to be explicit about the test by adding a skipif decorator to it.

[maxgalli]

The tests are passing, but this test, together with all the other ones that use xrootd_server https://github.com/scikit-hep/uproot5/blob/main/tests/conftest.py#L83 are being skipped.
This seems to happen in all the tests, and the reason for skipping is that XRootD is not installed. Is there a reason for not having XRootD as a dependency for the tests?

[maxgalli]

The tests are passing, but this test, together with all the other ones that use xrootd_server https://github.com/scikit-hep/uproot5/blob/main/tests/conftest.py#L83 are being skipped. This seems to happen in all the tests, and the reason for skipping is that XRootD is not installed. Is there a reason for not having XRootD as a dependency for the tests?

This happens for tests running on MacOS and Windows, the Ubuntu case is correctly tested. I opened an issue here #1384