Move sunpy.cm to sunpy.visualization.colormaps

[Cadair]

Fixes #3392

[ghost]

Thanks for the pull request @Cadair! Everything looks great!

[pep8speaks]

Hello @Cadair! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

• In the file sunpy/cm/__init__.py:

Line 7:101: E501 line too long (108 > 100 characters)
Line 12:1: E402 module level import not at top of file
Line 18:1: E402 module level import not at top of file
Line 19:1: E402 module level import not at top of file
Line 26:1: E402 module level import not at top of file

Comment last updated at 2019-10-30 18:01:46 UTC

[ayshih]

• Also update docs/dev_guide/sunpy_stability.yaml
• I suggest preserving the show_colormaps() figure in the code reference or docstring for sunpy.visualization.colormaps, unless there's a good reason not to

[ayshih]

See comments.

One more thing: it's beyond the scope of this PR, but it's fairly opaque how to use this package. The colormaps themselves are held in a list (cmlist), but this fact isn't documented anywhere. Are we intentionally avoiding providing access like sunpy.visualization.colormaps.sdoaia171?

[Cadair]

So another fun thing, we define an __all__ list in sunpy.visualization.colormaps.cm which excludes 99% of the objects in there, so only show_colormaps and cmlist (which isn't a list) get imported.

This is pretty much fine with the deprecated version, but what now doesn't work is this:

In [7]: from sunpy.cm.cm import sdoaia191                                                                                                                                                                                          
---------------------------------------------------------------------------
ModuleNotFoundError                       Traceback (most recent call last)
<ipython-input-7-c83856b488f1> in <module>
----> 1 from sunpy.cm.cm import sdoaia191

ModuleNotFoundError: No module named 'sunpy.cm.cm'

[ayshih]

Ha ha, of course, cmlist isn't a list. Awesome.

So, sunpy.cm.cm.sdoaia171 works, but from sunpy.cm.cm import sdoaia171 doesn't. Okay.

[Cadair]

Ha ha, of course, cmlist isn't a list. Awesome.

This might actually be a good time to change it.

So, sunpy.cm.cm.sdoaia171 works, but from sunpy.cm.cm import sdoaia171 doesn't. Okay.

Are we happy with this, because it did work before this change?

[Cadair]

Ok, with a little bit of high grade hackery, I fixed the import from issue. So if the CI agrees I think this is good to go.

[ayshih]

The figure-test build is still broken, though...

[nabobalis]

______________________________________________________________________ ERROR collecting sunpy/cm/__init__.py _______________________________________________________________________
sunpy/cm/__init__.py:9: in <module>
    category=_SDW)
E   sunpy.util.exceptions.SunpyDeprecationWarning: The functionality of the sunpy.cm module is now in sunpy.visualization.colormaps as of SunPy 1.1. The ability to import sunpy.cm as a module may be removed in a future version of SunPy.

Breaking on raising the DepWarn.

[ayshih]

Yeah, the open question during the dev meeting was why the figure-test build encounters that problem when collecting tests, but not, for example, the 32-bit build.

[ayshih]

Can't be merged until the issue with the figure-test build is resolved

[ayshih]

Technically, you're changing the behavior of from sunpy.cm import *, but that's because it didn't previously have an __all__. (For example, cmname would get imported.) However, the previous behavior is essentially a bug, and no one would have relied on it, so I support this change.

[Cadair]

I think it is only cname.

This __all__ is copied from sunpy.cm.cm.__all__ which we were importing with import *

[Cadair]

Can someone else please review this and then merge it

[dstansby]

😆