Section 22.3. Related Topics

22.3. Related Topics

SWIG

There is an external tool available called SWIG, which stands for Simplified Wrapper and Interface Generator. It was written by David Beazley, also the author of Python Essential Reference. It is a software tool that can take annotated C/C++ header files and generate wrapped code, ready to compile for Python, Tcl, and Perl. Using SWIG will free you from having to write the boilerplate code we've seen in this chapter. You only need to worry about coding the solution part of your project in C/C++. All you have to do is create your files in the SWIG format, and it will do the background work on your behalf. You can find out more information about SWIG from its main Web site located at the following Web address (URL):

http://swig.org

Pyrex

One obvious weakness of creating C/C++ extensions (raw or with SWIG) is that you have to write C/C++ (surprise, surprise), with all of its strengths, and, more importantly, its pitfalls. Pyrex gives you practically all of the gains of writing extensions but not the headache. Pyrex is a new language created specifically for writing Python extensions. It is a hybrid of C and Python, leaning much more toward Python: in fact, the Pyrex Web site goes as far as saying that, "Pyrex is Python with C data types." You only need to write code in the Pyrex syntax and run the Pyrex compiler on the source. Pyrex creates C files, which can then be compiled and used as a normal extension would. Some have sworn off C programming forever upon discovering Pyrex. You can get Pyrex at its home page:

http://cosc.canterbury.ac.nz/~greg/python/Pyrex

Psyco

Pyrex gives us the advantage that you no longer have to write pure C; however, do you need to learn its syntax, "yet another language." In the end, your Pyrex code turns into C anyway. You use C/C++, C/C++ with SWIG, or Pyrex because you still want that performance boost you are looking for. What if you can obtain performance gains without changing your Python code?

Psyco's concept is quite different from those other approaches. Rather than writing C code, why not just make your existing Python code run faster? Psyco serves as a just-in-time (JIT) compiler so you do not have to change to your source other than importing the Psyco module and telling it to start optimizing your code (during runtime).

Psyco can also profile your code to see where it can make the most significant improvements. You can even enable logging to see what Psyco does while optimizing your code. For more information, go to its main Web site:

http://psyco.sf.net

Embedding

Embedding is another feature available in Python. It is the inverse of an extension. Rather than taking C code and wrapping it into Python, you take a C application and wrap a Python interpreter inside it. This has the effect of giving a potentially large, monolithic, and perhaps rigid, proprietary, and/or mission-critical application the power of having an embedded Python interpreter. Once you have Python, well, it's like a whole new ball game.

For extension writer, there is a set of official docs that you should refer to for additional information.

Here are links to some of the Python documentation related to this chapter's topics:

Extending and Embedding

http://docs.python.org/ext

Python/C API

http://docs.python.org/api

Distributing Python Modules

http://docs.python.org/dist