Run Continuous Integration tests on Github Actions

[weiji14]

Description of proposed changes

Newer attempt of #35.

Fixes #

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]

The GitHub Actions look promising. It seems the macOS agent doesn't correctly cache the ~/.gmt directory (i.e., the cache size is only 22 B). Perhaps you could try to update the cache key to re-cache the directory?

[weiji14]

The GitHub Actions look promising.

Yeah, it's meant to consolidate all our Linux/macOS/Windows unit tests into one file using a matrix build specification (instead of having .travis.yml and .azure-pipelines.yml). Was going to put gmt-master into it somehow too at some point. It's been a lot of trial and error at this point getting it to work but should be getting there!

It seems the macOS agent doesn't correctly cache the ~/.gmt directory (i.e., the cache size is only 22 B). Perhaps you could try to update the cache key to re-cache the directory?

Good spotting, I'll try that. Not sure if it's because the Linux build runs faster than macOS and 'caches' things first.

[seisman]

GitHub Actions may not support $(HOME). Try ~/.gmt instead. See the pip cache example

[weiji14]

Ok, won't hurt to try.

[seisman]

Now macOS has the correct cache (~6MB), but Linux fails to extract the cache tarballs.

[weiji14]

Yeah. Not sure why the Windows run is saying it can't find the cache.

[weiji14]

I'll try using ${{ github.workspace }}/.gmt instead.

[seisman]

Still no luck. I noticed that Linux and Windows agents actually can download some files but timeout for others. For example:

Run gmt which -Gu @earth_relief_10m @earth_relief_60m @ridge.txt @Table_5_11.txt @tut_bathy.nc @tut_quakes.ngdc @tut_ship.xyz
earth_relief_10m: Download file from the GMT data server [data set size is 2.8M].
earth_relief_10m: Earth Relief at 10x10 arc minutes obtained by Gaussian Cartesian filtering (18 km fullwidth) of SRTM15+V2 [Tozer et al., 2019].

gmtwhich [ERROR]: Libcurl Error: Timeout was reached
gmtwhich [ERROR]: You can turn remote file download off by setting GMT_DATA_SERVER_LIMIT = 0.
gmtwhich [ERROR]: File earth_relief_10m.grd not found!
earth_relief_60m: Download file from the GMT data server [data set size is 106K].
earth_relief_60m: Earth Relief at 60x60 arc minutes obtained by Gaussian Cartesian filtering (111 km fullwidth) of SRTM15+V2 [Tozer et al., 2019].

C:/Users/runneradmin/.gmt/server/earth_relief_60m.grd
C:/Users/runneradmin/.gmt/cache/ridge.txt
C:/Users/runneradmin/.gmt/cache/Table_5_11.txt
C:/Users/runneradmin/.gmt/cache/tut_bathy.nc
C:/Users/runneradmin/.gmt/cache/tut_quakes.ngdc
C:/Users/runneradmin/.gmt/cache/tut_ship.xyz

I think a workaround would be running the "gmt which" command multiple times until all files are downloaded.

[ghost]

DeepCode failed to analyze this pull request

Something went wrong despite trying multiple times, sorry about that.
Please comment this pull request with "Retry DeepCode" to manually retry, or contact us so that a human can look into the issue.

[seisman]

Another typo:

Suggested change

	name: ${{ matrix.os }} - Python ${{ matrix.python_version }}
	name: ${{ matrix.os }} - Python ${{ matrix.python-version }}

[weiji14]

I see you've been working on it on a branch in upstream pygmt. Should I move this Pull Request there so we can both work on it?

[seisman]

Yes, that would be better.

[weiji14]

Closing in favour of upstream PR at GenericMappingTools#475.