[Visualpython-users] two directions

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

I just realized that there's a useful way of expressing the nature of the
conflicting views of needs and expectations of users. Arthur is
appropriately concerned about the needs and expectations of current Python
users who might consider adding the Visual module to their installation. I
have been concerned about the needs and expectations of nonusers of Python
(who are in many cases nonprogrammers) who are coming to Python (and even to
programming) for the first time. I can't begin to list all of the hidden
assumptions that are associated with thinking about these two opposite
directions of approach to VPython! A related way of capturing the two
directions is to say that the current Python user wants to add the Visual
module, not install VPython. "VPython" is a term that describes a bundle
that gets the newcomer to Python up and running and doing interesting things
quickly.

It was suggested that if I really wanted a simple installation I should have
bundled Python itself in with everything else, so that novices could do one
installation. As was asked, if two installers, why not three or more? If I
had seen how to make it one, I would have done it, because despite lots of
effort, even just two installations has caused problems (not frequent, but
annoying). Every additional installation will bring in its own chances for a
problem. In his "Brief History of Time," Stephen Hawking said that his
publisher told him that he would cut his readership in half by every
equation he included. He decided to include one equation and take the 50%
cut in audience.

On Windows, the current VPython installer uses Python registry entries to
find the PythonXX directory and install into it. Earlier installers merely
asked the user to install into the PythonXX directory, but many users didn't
and then couldn't figure out why nothing worked. Even with the
registry-based scheme I occasionally get reports from novices that they get
the message "can't find visual module". I've never been able to extract a
coherent statement of what went wrong, but if they uninstall and reinstall
they typically get it working, and I'm guessing that it has something to do
with where they installed Python. There's an echo here of a recent comment
by Guido in response to someone having a problem with Python when they chose
to install it somewhere other than in C:\PythonXX. His comment was basically
that they shouldn't have done that! So maybe there's something wrong with
Python itself with respect to location on Windows.

I'm hopeful that through the use of distutils we can make things good for
both approaches, either with appropriate interactions with the user in the
installation process or by providing two different packages.

Bruce Sherwood