(Rebooting) Python Usage Statistics

[Eric Snow]

The original topic got side-tracked pretty quickly (surprise,
surprise).  However, I think the question is valuable enough that it's
worth broaching again.  Notably, the first time around I did get a
meaningful response (thanks Stefan!).  If you do feel like diverging
from the topic of gathering usage statistics, please at least change
the subject.  <wink>

I've also started a wiki page:

  http://wiki.python.org/moin/usage_stats

Feel free to add to it.

--------------

Does anyone have (or know of) accurate totals and percentages on how
Python is used?  I'm particularly interested in the following
groupings:

- new development vs. stable code-bases
- categories (web, scripts, "big data", computation, etc.)
- "bare metal" vs. on top of some framework
- regional usage

I'm thinking about this partly because of a discussion on python-ideas
about [some U-word related topic <wink>].  All the rhetoric, anecdotal
evidence, and use-cases there have little meaning to me, in regards to
Python as a whole, without an understanding of who is actually
affected.

For instance, if frameworks (like django and numpy) could completely
hide the arguable challenges of [some feature] in Python 3--and most
projects were built on top of frameworks--then general efforts for
making [some feature] easier in Python 3 should go toward helping
framework writers.

Such usage numbers help us know where efforts could be focused in
general to make Python more powerful and easier to use where it's
already used extensively.  They can show us the areas that Python
isn't used much, thus exposing a targeted opportunity to change that.

Realistically, it's not entirely feasible to compile such information
at a comprehensive level, but even generally accurate numbers would be
a valuable resource.  If the numbers aren't out there, what would some
good approaches to discovering them?  Thanks!

-eric