fix #397 needed to store continent value

[doutriaux1]

No description provided.

[scottwittenburg]

Interesting, I guess while I knew you could disable the continents somehow, I never knew this is how you did it.

[scottwittenburg]

After browsing the source code, I'm not seeing how this value stored on the display is ever queried to avoid plotting the continents. To me it looks like the decision is made inside our vcs2vtk.genGrid(...) function, based solely on the type of data we get. But I guess it's working, so it's fine with me.

[doutriaux1]

you're right, it's sort of obscure but i think it's there: https://github.com/CDAT/vcs/blob/master/vcs/VTKPlots.py#L491

[scottwittenburg]

Ah, you referred to it with an underscore in your fix, so I was searching for it that way, forgetting about how you typically make "properties" out of those attributes. Thanks.

[doutriaux1]

@scottwittenburg failures are in doc build/sphinx I think i'ts safe to merge from that point of view, doesn't seem related to this PR

[scottwittenburg]

I was just searching through the CI output for "error", there were 86 occurrences, making it hard to tell what's going on. Some were image comparison size mismatches:

Executing nosetests -A not vtk_ui -s /Users/distiller/project/tests/test_vcs_dejavu_fonts.py in /Users/distiller/project
b'ERROR: In ../Imaging/Core/vtkImageDifference.cxx, line 558'
b'vtkImageDifference (0x7fd2aa9c7ad0): ExecuteInformation: Input are not the same size.'
b' Input1 is: 0,813,0,302,0,0'
b' Input2 is: 0,813,0,303,0,0'

But there are a bunch of those in the passing py2 check, but what the heck? Why don't those cause test failures? Whatever. Then there are a bunch similar to this:

CDMS error: Error on relative units conversion, string =

And again, many of the same in the py2 case. I have no idea what that's about. In both cases (check passed vs check failed), there are also some errors I've seen before coming from VTK when we do projection and an input point is outside the range it can handle:

Generic Warning: In ../Common/Core/vtkMath.cxx, line 743
vtkMath::Jacobi: Error extracting eigenfunctions

From what I can tell, though, it seems you're right that the only kind of error we see in the failing case which we don't also see in the passing case, is this one:

Extension error:
Could not import extension sphinx.builders.epub3 (exception: No module named 'sphinxcontrib')
make: *** [doctest] Error 2
Exited with code 2

Has that been happening with other PRs recently?

[coveralls]

Coverage remained the same at ?% when pulling d2f70be on issue_397_update_brings_back_continents into e607721 on master.