Intro to writing C-extension for Python

http://danfm.ca/posts/python-c-extensions/

by synparb • 14 years ago

89 7 14 years ago

danfm.ca

Also has some nice numpy specific stuff for scientific code

synparb• 14 years ago

This piece also has some nice numpy specific information for scientific computing.

zaptheimpaler• 14 years ago

This and Bitey(https://github.com/dabeaz/bitey) are both on the front page :)

thepumpkin1979• 14 years ago

omg, I suck at math. am I the only one who got intimidated by the math formula? May be I need a tutorial about C ABI/Python writing wrappers like "foo_sum" and "bar_multi".

vog• 14 years ago

That's not math. It's just a formula. It is nothing more than the C function, written in a more compact notation. Consider this notation as some kind of programming language.

Learning the meaning of math symbols is actually easier than learning a new programming language, or some cumbersome API. (I assume that you have no problem picking up a new language, as this article is directed to people who have learned at least two quite different languages: C and Python.)

So if it's just a formula, what would be math? Math would be describing exactly what problem is solved by this function, proving that the formula's result will indeed have the desired properties, and checking all edge cases and interesting special cases of that function. And on top of all this, math would analyze the effects of floating point inaccuracies on the result.

So this article doesn't really contain math. It contains just a formula, which shouldn't be any more intimidating than the corresponding C function.

16s• 14 years ago

Boost Python is great for anyone doing C++ Python modules. It's rather simple too.

LBarret• 14 years ago

but slow.

jnazario• 14 years ago

FWIW you could have done this in far fewer lines of python using a code generator like Cython or Pyrex.

that said, learning the intricacies of an extension - and all of the stuff you have to do under the hood like registering handlers, reference counting, cleanup, etc - is a valuable experience.

but once you got that done i just recommend you use code generators. it's also a lot easier to maintain.

Intro to writing C-extension for Python

Related Stories