PyCon X


2nd - 5th May 2019

Simple APIs and innovative documentation processes: looking back at the success of Scientific Python

A large and growing fraction of scientists enthusiastically advocate the use of Python, a general-purpose language. In this talk, I will highlight two essential factors of this success. First, the NumPy numerical array object provides a solid foundation on which scientific packages have built simple (and yet powerful) APIs. As I will show from the examples of scikit-learn and scikit-image, such chiseled APIs have shrunk down to a minimal form, thus reducing greatly the energy barrier to use the libraries and improving users’ code maintainability and readability. Second, I will present several innovative initiatives that helped developing several complementary forms of user documentation since the mid 2000’s, such as the collaborative definition of a documentation standard, documentation marathons, or the more recent graphical example galleries. I will illustrate this talk with several examples from the use and development principles of scikit-image, an image processing library relying on NumPy arrays, of which I’m a core contributor.

Do you have some questions on this talk?

New comment