Howdy Xavier!
[Apologies for the length of this; I didn't expect to write so much!]
I've been a Python programmer for many years now (having come from a
PHP, Perl, C, and Pascal background) and I'm constantly learning new
idioms and ways of doing things that are more "Pythonic"; cleaner, more
efficient, or simply more beautiful. I learn by coding, rather than by
reading books, taking lectures, or sitting idly watching screencasts.
I constantly try to break the problems I come up with in my head into
smaller and smaller pieces, then write the software for those pieces in
as elegant a method as possible.
Because of my "turtles all the way down" design philosophy, a lot of my
spare time projects have no immediate demonstrable benefit; I code them
for fun! I have a folder full of hundreds of these little projects,
the vast majority of which never see a public release. I also collect
little snippets of code that I come across[1] or write, and often
experiment with performance tests[2] of small Python snippets.
Often I'll assign myself the task of doing something far outside my
comfort zone; a recent example is writing a HTTP/1.1 web server. I had
no idea how to do low-level socket programming in Python, let alone how
HTTP actually worked under-the-hood, and because my goal wasn't
(originally) to produce a production-quality product for others it gave
me the freedom to experiment, rewrite, and break things in as many ways
as I wanted. :) I had people trying to convince me that I shouldn't
re-invent the wheel ("just use Twisted!") though they mis-understood
the reason for my re-invention: to learn.
It started as a toy 20-line script to dump a static HTTP/1.0 response
on each request and has grown into a ~270 line fully HTTP/1.1
compliant, ultra-performant multi-process HTTP server rivalling pretty
much every other pure-Python web server I've tested. (I still don't
consider it production ready, though.) Progressive enhancement as I
came up with and implemented ideas meant that sometimes I had to
rewrite it from scratch, but I'm quite proud of the result and have
learned far more than I expected in the process.
While I don't necessarily study books on Python, I did reference HTTP:
The Definitive Guide and many websites in developing that server, and I
often use the Python Quick Reference[3] when I zone out and forget
something basic or need to find something more advanced.
In terms of understanding how Python works, or how you can use certain
semantics (or even better, why you'd want to!) Python Enhancement
Proposals (PEPs) can be an invaluable resource. For example, PEP
318[4] defines what a decorator is, why they're useful, how they work,
and how you can write your own. Pretty much everything built into
Python after Python 2.0 was first described, reasoned, and discussed in
a PEP.
If you haven't seen this already, the Zen of Python[5] (a PEP) has many
great guidelines. I try to live and breathe the Zen.
So that's my story: how I learn to improve my own code. My motto,
"re-inventing the wheel, every time," is the short version of the
above. Of course, for commercial work I don't generally spend so much
time on the nitty-gritty details; existing libraries are there for a
reason, and, most of the time, Getting Things Done™ is more important
than linguistic purity! ;)
— Alice.
[1] https://github.com/GothAlice/Random/
[2] https://gist.github.com/405354
[3] http://rgruet.free.fr/PQR26/PQR2.6.html
[4] http://www.python.org/dev/peps/pep-0318/
[5] http://www.python.org/dev/peps/pep-0020/
--
http://mail.python.org/mailman/listinfo/python-list
