Improve warped rectangular projections #1180
Conversation
These tests all appear in most of the other projections' tests.
QuLogic
force-pushed
the
improve-warped
branch
from
7e77064
to
08982cc
Compare
|
Also, instead of using existing results as I did before, I worked out the limits from the equations (that's how I noticed Mollweide didn't support ellipses.) |
| @@ -175,9 +175,11 @@ def test_transform_points_xyz(self): | |||
| def test_globe(self): | |||
There was a problem hiding this comment.
Interesting that this test hasn't changed based on the fact that the globe changed. Shows that the test is of questionable value. Perhaps the first thing to fix is the name of the variables (rugby_globe -> footy_globe and footy_globe -> tennisball_globe). And then we should think about whether this test needs to remain at all...
Thoughts?
There was a problem hiding this comment.
I mean, that's kind of the point; Mollweide doesn't support elliptical globes, so specifying ellipse or semiminor_axis does nothing.
This test doesn't actually check that the results are different in the end... only that they match some 'true' value, whatever that might be.
There was a problem hiding this comment.
Oh, perhaps I was a bit unclear; that's the point of the changes in this file. As to the point of the test, I think it's just testing that the globe does something, though maybe it's become a bit redundant by the individual test_eccentric_ellipse tests in the crs directory.
Maybe we should add a test_globe for the ones that only support spheres and remove this test?
There was a problem hiding this comment.
Maybe we should add a test_globe for the ones that only support spheres and remove this test?
Either that, or we keep the test but fix the variable name. Either works for me. I'll defer this though in order to move the PR forwards.
| assert np.all(np.isnan(result)) | ||
|
|
||
|
|
||
| def test_transform_points(): | ||
| """ | ||
| Mostly tests the workaround for a specific problem. | ||
| Problem report in: https://github.com/SciTools/cartopy/issues/23 |
There was a problem hiding this comment.
Missing a 2 at the end (issue 232, I believe)
|
Oh, I might have been wrong about Mollweide; it seems to make a difference using WGS84 spherical-ized semimajor radius vs. WGS84-the-ellipse. But I'm not sure exactly if it truly supports ellipses. |
|
Oops, copy&paste type; was comparing against Robinson with the ellipse. Never mind. |
Improves warped rectangular projections in a few ways: