A friend asks,

i'm going to be recoding <x> from scratch starting next week, in python.
what codebase would you recommend as good to model after?

Any thoughts on a well-formed, reasonably sized (yet not huge), and simple Python code base?

There have to be some examples somewhere! I'd suggest something in the stdlib, but nothing is coming to mind right now -- and there are some real stinkers in there, too.

For now, I've pointed my friend towards PEP 8 and the Python Cookbook. Very unsatisfying.

--titus

---

Legacy Comments

Posted by Greg Wilson on 2009-01-07 at 15:54.

If you find one, please let me know --- my students would want to see
it too.

Posted by Joseph Lisee on 2009-01-07 at 17:03.

Trac: <a href="http://trac.edgewall.org/browser/trunk/trac">http://tra
c.edgewall.org/browser/trunk/trac</a>    Its on the bigger size, but
they use a very neat extension point and plugin system which makes
modifying and expanding Trac very easy and powerful.  I built a
similar plugin/extension point system myself based on some of the
"magic" Trac uses to make there's work.  Trac also has decent package
organization and testing.

Posted by Brett on 2009-01-07 at 17:13.

Typically the newer modules in the standard library that were
specifically for the stdlib and not imported from external use tend to
be very good. I remember thinking at one point that the heapq module
was of high quality, for instance.

Posted by Jonathan Ellis on 2009-01-07 at 17:18.

I guess your objection to the Cookbook (dead trees version) is that
the projects are too small?  Other than that I thought the commentary
was quite good.

Posted by Titus Brown on 2009-01-07 at 17:31.

Jonathan -- exactly.    Joseph, Trac is big enough that I worry it's
got its whole own set of conventions.  It's definitely hard to grok.
I guess I'm looking for something in the "sweet spot" between a single
module and a large-ish program: small enough to understand, big enough
to have to confront some real problems.    I'd offer up twill, but
it's just a wrapper around mechanize, really.

Posted by hads on 2009-01-07 at 17:35.

Check out Zine; <a
href="http://zine.pocoo.org/">http://zine.pocoo.org/</a> - it's a
pretty new project but the guys behind it write nice code in my
opinion.

Posted by Rene Dudfield on 2009-01-07 at 17:47.

hi,    great question.    I wouldn't recommend the stdlib modules...
often they are only written for the latest version of python - and are
made for distribution with python - not for external distribution.
There are ***many*** things you need to do for high quality external
packages that you don't need to do with stdlib modules.    Good python
modules often need to work with older releases of python.    You'll
probably need to make heaps of hacks/extensions to distutils, have
automated bots building and testing it all the time, plus all sorts of
other things.    You need to have it packagable for the various OS
distributions, like debian, freebsd, macports, windows installers, mac
OSX installers etc etc.  Not to mention setting it up in pypi,
easy_install, and buildout.    A good small software needs to have it
all in one file with tests and all :)  Then people can just copy that
file into their project and they are done.    Then there is building a
community around the project... which is often very important.  So
there are lots of examples available, and posts about how to do
things.  Can people easily contribute?  Can people test the software
on their platform easily?    There is also promotion -- where to
announce the project?  Documentation about why the software is useful,
it's pros and cons.    Also, applications have many different
requirements compared to frameworks, or libraries.    I guess it
really depends on the type, and size of the software, and the
distribution needs.    For big projects, I'd look at cherrypy,
twisted, pygame, moinmoin, trac, and numpy.  Each project has
different requirements and needs.  I'd call each of these projects
good for their requirements.      ... anyway, there's some rambling
thoughts on the matter for you... I'm very interested in what others
have to say about the question.

Posted by Giulio Piancastelli on 2009-01-07 at 19:35.

I'm in a similar position as your friend. I had a look at html5lib
(http://code.google.com/p/html5lib/): I found some inspiring bits, but
I am really a newbie at Python coding and code base structuring, so
I'm not in the best position to judge html5lib's quality.    If you
look at it, I'd be interested in reading your thoughts.

Posted by Ben on 2009-01-08 at 07:03.

web.py

Posted by Tom on 2009-01-08 at 19:55.

It is by no means small, but one of the best I've seen is Twisted.
Religious about TDD, automated builds, etc. The codebase is pretty
enormous, but you could probably showcase any individual module for a
smaller slice.    <a
href="http://twistedmatrix.com">http://twistedmatrix.com</a>

Posted by Chris Perkins on 2009-01-09 at 19:15.

I think the sqlalchemy codebase is a work of art.    Twisted codebase
is good, but it does not follow pep-8.    The Pylons codebase is nice,
as are Formencode, and PasteScript.

Posted by Yi Qiang on 2009-01-10 at 23:38.

Another vote for Twisted, some of the best code I've had a chance to
work with yet.

Posted by Yi Qiang on 2009-01-10 at 23:38.

Another vote for Twisted, some of the best code I've had a chance to
work with yet.