shapely dependence should not be a requirement

[emorway-usgs]

I recently started getting the message below when trying to import flopy If the user is not intending to make use of the gridintersect.py class, should this dependence be a requirement as it now seems to be?

[jwhite-usgs]

On the develop branch, I think the shapely deps are wrapped in a try-except. Are you using develop?

[emorway-usgs]

Affirmative, on the develop branch

[jwhite-usgs]

python 2? Looks like there is some trapping for py2 right above the shapely imports

[emorway-usgs]

python 3.5.3 (see top of screen shot in first post of thread)

[jwhite-usgs]

alright, Im giving up then!

[christianlangevin]

I've seen this too. @jwhite-usgs, remember when you added the import of gridintersect to the __init__.py file? At that point, I think it became a requirement, because then it is imported when you import flopy. We probably need a try and except in the gridintersect.py file.

[jwhite-usgs]

there is a try-except but just before that, there is a re-def of importerror to some other form of exception if python=2. I think that logic might be what we are seeing here

[christianlangevin]

Right, and ModuleNotFoundError isn't defined otherwise.

[christianlangevin]

Should probably put the imports into the methods of gridintersect.GridIntersect() that need them.