Support passing data in numpy int8, int16, uint8 and uint16 dtypes to GMT

[seisman]

Description of proposed changes

Support numpy data types np.int8, np.int16, np.uint8, and np.uint16.

References:

• https://numpy.org/doc/stable/reference/arrays.scalars.html?highlight=uint32#sized-aliases
• https://github.com/GenericMappingTools/gmt/blob/b552be3a3f76455cd45f32202d468d14ce93cd28/src/gmt_resources.h#L83

Fixes #1960.

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 wrapping a new module, open a 'Wrap new GMT module' issue and submit reasonably-sized PRs.
• If adding new functionality, add an example to docstrings or tutorials.

Slash Commands

You can write slash commands (/command) in the first line of a comment to perform
specific operations. Supported slash commands are:

• /format: automatically format and lint the code
• /test-gmt-dev: run full tests on the latest GMT development version

[weiji14]

Awesome @seisman! Didn't realize GMT even supported these other dtypes!

Just need to update some tests, e.g. these lines:

pygmt/pygmt/tests/test_clib_put_matrix.py

Line 18 in 1067fa3

dtypes = "float32 float64 int32 int64 uint32 uint64".split()

pygmt/pygmt/tests/test_clib_put_matrix.py

Line 64 in 1067fa3

dtypes = "float32 float64 int32 int64 uint32 uint64".split()

pygmt/pygmt/tests/test_clib_put_vector.py

Line 19 in 1067fa3

dtypes = "float32 float64 int32 int64 uint32 uint64".split()

pygmt/pygmt/tests/test_clib_put_vector.py

Line 59 in 1067fa3

dtypes = "float32 float64 int32 int64 uint32 uint64".split()

Also 6 more lines in test_clib.py. Maybe a good idea to make the dtypes list into a pytest fixture, use something like:

Use dtypes = "int8 int16 int32 int64 uint8 uint16 uint32 uint64 float32 float64".split()

[seisman]

Awesome @seisman! Didn't realize GMT even supported these other dtypes!

Just need to update some tests, e.g. these lines:

pygmt/pygmt/tests/test_clib_put_matrix.py

Line 18 in 1067fa3

dtypes = "float32 float64 int32 int64 uint32 uint64".split()

pygmt/pygmt/tests/test_clib_put_matrix.py

Line 64 in 1067fa3

dtypes = "float32 float64 int32 int64 uint32 uint64".split()

pygmt/pygmt/tests/test_clib_put_vector.py

Line 19 in 1067fa3

dtypes = "float32 float64 int32 int64 uint32 uint64".split()

pygmt/pygmt/tests/test_clib_put_vector.py

Line 59 in 1067fa3

dtypes = "float32 float64 int32 int64 uint32 uint64".split()

Also 6 more lines in test_clib.py. Maybe a good idea to make the dtypes list into a pytest fixture, use something like:

Use dtypes = "int8 int16 int32 int64 uint8 uint16 uint32 uint64 float32 float64".split()

Was working on it. Now pushed to this branch.

[seisman]

I just realized that passing str_ type matrix to GMT is allowed in the PyGMT put_matrix function, but was not documented here. But I'm wondering what happens if passing a str matrix to GMT.

[weiji14]

Suggested change

	Not at all numpy dtypes are supported, only: int8, int16, int32, int64,
	uint8, uint16, uint32, uint64, float32, float64, str\_ and datetime64.
	Not all numpy dtypes are supported, only: int8, int16, int32, int64,
	uint8, uint16, uint32, uint64, float32, float64, str\_ and datetime64.

I just realized that passing str_ type matrix to GMT is allowed in the PyGMT put_matrix function, but was not documented here. But I'm wondering what happens if passing a str matrix to GMT.

Maybe try and see?

[weiji14]

Suggested change

	Not at all numpy dtypes are supported, only: int8, int16, int32, int64,
	uint8, uint16, uint32, uint64, float32, float64, str\_ and datetime64.
	Not all numpy dtypes are supported, only: int8, int16, int32, int64,
	uint8, uint16, uint32, uint64, float32, float64, str\_ and datetime64.

I just realized that passing str_ type matrix to GMT is allowed in the PyGMT put_matrix function, but was not documented here. But I'm wondering what happens if passing a str matrix to GMT.

Maybe try and see?

[weiji14]

Just as an aside, is there any chance of GMT supporting half precision floats? I.e. https://numpy.org/doc/stable/reference/arrays.scalars.html#numpy.float16, or 2-byte floats. Not sure if it's possible in C, but supporting float16 will be useful for some machine learning workloads that use float16.

[seisman]

Looking at the source code of the GMT_Put_Matrix function (https://github.com/GenericMappingTools/gmt/blob/b552be3a3f76455cd45f32202d468d14ce93cd28/src/gmt_api.c#L15458) and the gmtapi_insert_vector function (https://github.com/GenericMappingTools/gmt/blob/b552be3a3f76455cd45f32202d468d14ce93cd28/src/gmt_api.c#L15297), I think only numeric data types are supported.

[weiji14]

Looks ok to me, but maybe wait and see if @De-Guo has time to verify that it fixes his issue at #1960.

[seisman]

Just as an aside, is there any chance of GMT supporting half precision floats? I.e. numpy.org/doc/stable/reference/arrays.scalars.html#numpy.float16, or 2-byte floats. Not sure if it's possible in C, but supporting float16 will be useful for some machine learning workloads that use float16.

I have no idea about float16. Better to open a feature request in the GMT repository.