About reading Python code
stargaming at gmail.com
Mon Mar 17 06:54:42 CET 2008
On Sun, 16 Mar 2008 20:54:01 -0700, WaterWalk wrote:
> Hello. I wonder what's the effective way of figuring out how a piece of
> python code works.
If your Python code is well-written, it should be easy figuring out what
it means by just reading it. For more complex programs, of course, this
method can fail.
> With C I often find it very useful to be able to run
> the code in step mode and set breakpoints in a debugger so I can watch
> how the it executes, how the data change and how the code jumps from one
> function to another. But with Python, the debugger is a little
> primitive. The default IDLE doesn't even allow me to set a breakpoint.
> When the code is long, I am often lost in it.
IDLE is just, well, a batteries-included editor. There are many people
(me included) who do *not* like it because it's so weak and doesn't have
any real uses cases if your programs get sufficiently complex (because
IDLE itself is sufficiently primitive).
You might be interested in *real* debuggers, such as the Python Debugger
`PDB <http://docs.python.org/lib/module-pdb.html>`_. If you don't find
its usage obvious, a quick google search just turned up a nice `tutorial
You might find the `Python Profilers <http://docs.python.org/lib/
profile.html>`_ particularly interesting, `profile` for finding out which
function sucks up the most calls/time, `trace` for more sophisticated
`pythontracer <http://code.google.com/p/pythontracer/>` sounds like a
good combination of both.
> So I'm curious how to read code effectively. I agree that python code is
> clear, but when it becomes long, reading it can still be a hard work.
A common practice is just inserting `print` statements since it's so
easy. If you think your debugging isn't temporary but could be useful and
will be enabled every now and then, you could also use the `logging
module <http://docs.python.org/lib/module-logging.html>`_ with the
There was a blog post recently about how to do this `generically for
> BTW. I think this problem also exists in other script languages.
FWIW, I think it's particularly easier in scripting languages to
implement all kinds of tracing (apart from debugging, which is rather
unpopular in Python) because you have one *extra* level (the interpreter)
between your machine and your code.
More information about the Python-list