[python-advocacy] Obstacles to the adoption of Python

[Jeff Rush]

Rex Eastbourne wrote:
> 
> I am new to this list. I am a longtime Python user and like it quite a 
> bit; I am interested in getting involved in advocacy.

Welcome Rex!


> I've been thinking recently about the best way to increase the adoption 
> rate of Python. I suppose one valuable thing to do would be to do 
> research on the reasons -- real or imagined -- that people are *not* 
> adopting it. For instance, I saw the following blog post yesterday:
> 
> http://coffeeghost.net/2008/03/19/your-ignorance-does-not-make-a-programming-language-suck/
> 
> Has anyone done an analysis of the main practical issues holding Python 
> back?

Not that I know of, in a rigorous way.  As an advocate of Python, I have in my 
head one high-level map of its strengths and weaknesses.


 > Off the top of my head, here are two commonly cited obstacles:
> 
> -Lack of a popular, mature add-on repository (I know there is PyPI, but 
> it doesn't seem to be as mature as CPAN)

At PyCon that wrapped up this week, this issue has been elevated and I and 
others with experience in repositories outside the Python community are just 
starting to lay the groundwork for major improvements.  I agree that the ease 
of identifying cool software, of grabbing it, along with the necessary 
dependencies and installing with minimal hassle is an important factor for 
language adoption.  If you want to get involved, join the distutils-sig 
mailing list:

    http://mail.python.org/mailman/listinfo/distutils-sig

> -Documentation (I have no big problem with it, but many people feel the 
> online Python documentation is not up to par with what other languages 
> have, like perldoc)

Documentation could use work in a variety of areas.  We're moving from 2.5 
LaTeX based docs to 2.6 ReST based docs, with a much nicer website.  Compare:

    http://docs.python.org/

to the newer:

    http://docs.python.org/dev/

However, there are still modules that are undocumented or underdocumented. 
There are many pages that need actual examples that people can clip and use.

You can see here, the raw content,

    http://svn.python.org/view/python/trunk/Doc/

suitable for grabbing, patching and submitting for review, here:

    http://bugs.python.org

if you don't have commit privileges directly.

We need high-level guides to help in choosing the correct module for the job 
at hand.  Having ten tools that do something slightly different is great for 
the expert who knows why, but the novices need explanations of how to choose. 
  We need guides on testing methodologies, web and GUI framework selection, 
monetary currency issues (decimal + locale + Python cookbook), concurrent 
programming tradeoffs (threads, processes, event-based, GIL discussion), 
database tradeoffs (RDBMS, ZODB/Durus, pickle/shelf).

Some aspects of it are at:

     http://www.python.org/about/apps/

It'd be cool if the docs were is such a form that, when installed onto a local 
machine, the act of installing packages caused additional chapters to appear 
in the manuals.  This is an interplay between Python packaging and Python 
documentation projects.

There is no agreement in the markup to use within docstrings, although there 
are several competing styles, with most people not using any, hindering the 
automatic generation of common docs beyond basic introspection metadata.

> Specifically, it is "logistical" issues like the above that interest me 
> more than the technical issues (execution speed, language syntax) that 
> seem to be getting most of the attention. For one, these logistical 
> issues can be more straightforward to implement.

More straightforward... ;-) the logistical ones have more politics and are 
less objective, making it harder to move forward.  But often there is not 
resistance so much as a lack of volunteer energy and "coming together". 
Python's popularity means so many of us are very busy.


> Does anyone have stuff to add to this list?

A confusing array of multimedia options, both in the standard library and 
outside, with varying qualities of cross-platform integration and unclear 
tradeoffs in realtime performance, flexibility and licensing.

A need for a well-written guide on how to debug and profile your Python 
programs - one of the most common requests from newbies on various lists and 
screencast sites.  Bug of the Week, Find the Ten Bugs in This Program.

A lack of a security sandbox, similar to Java.  It prevents the use of Python 
as a first-class scripting language in browsers, makes difficult the use of 
Crunchy teaching software and drags some on thru-the-web webframeworks.  There 
is a prototype capability-based solution that is cool, but it needs community 
involvement.

   http://mail.python.org/pipermail/python-dev/2006-September/068673.html

No searchable, maintained place, w/reviews, where to buy, in/out print, for 
someone to locate a book about Python appropriate to their skill level and 
area of use.  Bonus if it can be sub-queried such that various Python teams 
like Django and Zope can embed it on their websites and provide bracketed book 
location service for just their interests.

No geocoded, searchable registry of Python usergroups, speakers or consultants.

See anything you'd like to get involved in? ;-)

-Jeff