Add a test for xarray shading

[seisman]

Description of proposed changes

This PR adds a test checking if grid shading works for xarray.

It may work for GMT 6.1.0, but unfortunately, it doesn't work for GMT 6.1.1.

Tried to address #364 but failed.

Reminders

• Run make format and make check to make sure the code follows the style guide.
• Add tests for new features or tests that would have caught the bug that you're fixing.
• Add new public functions/methods/classes to doc/api/index.rst.
• Write detailed docstrings for all functions/methods.
• If adding new functionality, add an example to docstrings or tutorials.

[seisman]

Here are the generated images:

Using grid	Using xarray	GMT CLI (no shading)

The GMT CLI one is generated using the following command:

gmt grdimage @earth_relief_01d_g -R-180/180/-90/90 -B -JCyl_stere/6i -Cgeo -png test_grdimage_xarray_shading-noshading

It seems shading is not applied for xarray grids.

[weiji14]

Weird, can you try with shading="+d" as @liamtoney mentioned at #364 (comment)?

[seisman]

No changes at all.

[seisman]

In #364 (comment), I mentioned that shading works well for 6.1.0, but now no shading at all, and sometimes crashes.

[weiji14]

Possible regression in GMT? 🤦 Hold on, let me try this branch.

[seisman]

Sometimes I even get an image like this one:

[seisman]

Possible regression in GMT?

Yes, that's possible. There are some bigger changes in dealing with matrix grids in 6.1.1.

[weiji14]

Yes, I can confirm this locally. Also get an whitened image sometimes (compared to your darker one). Next thing I would try is with the GMT_IS_DUPLICATE/GMT_IS_REFERENCE settings.

[seisman]

The test won't pass unless the upstream GMT fixes the bug. Maybe we can mark the test as "xfail" and merge it?

[weiji14]

The test won't pass unless the upstream GMT fixes the bug. Maybe we can mark the test as "xfail" and merge it?

Do we know what's wrong here (since 6.1.0 works)? I'm ok with using xfail, but please remove "Fixes #364." from the top post first.

[seisman]

Do we know what's wrong here (since 6.1.0 works)?

There are a lot of GMT API changes in GMT 6.1.1. These changes fix some crashes and issues, but unfortunately, break the grid shading. We may file another bug report to GMT later, and hopefully, it can be fixed again in GMT 6.2.0.

[weiji14]

Can we take this PR out of draft mode @seisman? It looks ready for review/approval, the Windows test failure is just that random one that occasionally breaks.