Re: [Visualpython-users] Let's hash out the documentation for restrictive imports

[Bruce S. <bas...@nc...>]

In a private conversation Craig Struble has helped me see another
structure that may be particularly appropriate. I believe his thinking
is in part based on Guy's suggestions.

Craig suggests implementing "a style of usage similar to the pylab
module, where the really pythonish stuff is all in matplotlib.foo
(where foo is some specific part of the API). A motivation for this is
that I believe the community of folks using matplotlib, numpy, and
scipy are potential users of visual. They will be familiar with the
matplotlib model." In site-packages there is a single file pylab.py
whose import gives you everything, and there's a folder matplotlib
containing the stuff that pylab.py imports. Mimicking this, there
would be a single file visual.py whose import gives you everything,
and a folder (maybe named "vpython") that would contain all the
submodules.

I pointed out that while we can arrange things in such a way as to
look like this, the actual implementation has to be "the other way
round": for backward compatibility the visual folder has to remain
essentially unchanged, and a second folder (maybe named "vpython")
would have submodules that would import from the visual folder. The
reason for this is that people may be importing from
visual.primitives, and they are certainly importing from visual.graph.

Bruce Sherwood

P.S. Incidentally, in the new vpython folder there should be a new
version of the graph module, because the existing one executes the
dreaded "from visual import *", and changing that in visual.graph
would cause problems, because people have been able to say just "from
visual.graph import *" and get spheres as well as graphs. The new
version, in the vpython folder, of course would not import spheres
etc.