Fix logic for no geoviews explorer

[ahuang11]

Even though geo=False, explorer would raise raise ImportError('GeoViews must be installed to enable the geographic options.')

import pandas as pd
import hvplot.pandas
df = pd.DataFrame({'a': [1,2,3], 'b': [4,5,6]})

df.hvplot.explorer(geo=False)

Not sure how to add a test for this because the dev dependencies installs geoviews.

[hoxbro]

For the test you could try https://github.com/holoviz/holoviews/blob/4a9e60990bd287e0f39b24d569f65b1ae459d4a0/holoviews/tests/conftest.py#L101

[ahuang11]

ooh that's fancy and good to know! Turns out there actually was a marker for geo tests so I didn't need it.

[codecov]

Codecov Report

Attention: Patch coverage is 90.47619% with 2 lines in your changes missing coverage. Please review.

Project coverage is 89.03%. Comparing base (9078378) to head (70e59b3).
Report is 7 commits behind head on main.

Files with missing lines	Patch %	Lines
hvplot/util.py	71.42%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1451      +/-   ##
==========================================
+ Coverage   88.94%   89.03%   +0.09%     
==========================================
  Files          52       52              
  Lines        7781     7819      +38     
==========================================
+ Hits         6921     6962      +41     
+ Misses        860      857       -3     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[hoxbro]

Are you sure?

[maximlt]

ooh that's fancy and good to know! Turns

It's also a little crazy right 🙃

A suggestion could be to refactor gv_available into a utility function that is mocked for this test to return False.

[ahuang11]

Hm I guess there is a marker, but the tests still run everything
https://github.com/holoviz/hvplot/blob/main/.github/workflows/test.yaml#L171
https://github.com/holoviz/hvplot/blob/main/hvplot/tests/testui.py#L394

I guess I'll go with Maxime's suggestion

[maximlt]

I got confused a minute here, does find_spec('geoviews') actually import GeoViews?

[hoxbro]

No, but it checks if it is installed.

[maximlt]

Yes ok that's what I thought. I was confused as then an ImportError is raised. Since importing GeoViews comes with side effects (registering lots of stuff), I'd like to make sure we're not removing an important step there. @hoxbro did you suggest find_spec for performance reasons?

[hoxbro]

Just as an alternative and also easier to patch in the test suite.

I think it is okay to use find_spec when we check if it is available. We should import geoviews in the code, where there previously was a try/except, to get those juicy side effects.

[hoxbro]

I also think the error message is wrong here. As it no longer only affects the UI code.

[maximlt]

@ahuang11 what do you think about the refactor done in 31df383? I'm fine with reverting if you prefer the former way. I've preferred creating a function that does the actual import since the side effects of importing geoviews as required I think.

[ahuang11]

I'm fine with the refactor.