January 29, 2012

Python - Matplotlib and Numpy on Debian/Ubuntu

There are `python-matplotlib` and `python-numpy` packages in the Debian/Ubuntu repos.

However, if you want to run in a virtualenv (with no-site-packages), and pip install these packages from PyPI, you need some system dependencies installed first to build with:

$ sudo apt-get install build-essential python-dev libfreetype6-dev libpng-dev python-virtualenv

Then, you can create a virtualenv, and the installers for Numpy and Matplotlib will work:

$ virtualenv env
$ cd env
$ source bin/activate
(env)$ pip install numpy matplotlib

Successfully installed numpy matplotlib
Cleaning up...

(tested on Ubuntu Oneiric 11.10 and Ubuntu Precise 12.04 alpha)

January 1, 2012

Officially Introducing "SST" (Python Web Test Framework)

"SST (selenium-simple-test) is a framework built on Selenium WebDriver, using Python to make writing functional web tests easier with code."

Since early 2011, I have been working for Canonical on the Infrastructure Systems Development team (Core Dev Ops).

[pic of canonical-isd-hackers at UDS-P-Orlando]

A by-product of our recent development efforts is a web testing framework. It has been available on Launchpad for a while, but I've never really announced it in public. We are using SST internally, and I want to expose it to a wider audience.

What is Selenium WebDriver?

Selenium (WebDriver) is a popular open-source library for automating browsers. It can be used to create functional/acceptance tests of a web application. The Selenium client bindings provide API's that allow you to programatically drive a browser and access web content/elements. The bindings are available and supported for many languages and platforms.

While working directly with Selenium API's from code is fine for ad-hoc browser interaction, it is rather low-level and lacks things necessary for creating suites of automated web tests. For larger-scale testing, you will soon want to use a framework to help organize, execute, and report.

Introducing SST...

SST Logo

SST aims to keep things simple.

Tests are made up of scripts, created by composing actions that drive a browser and assert conditions. You have the flexibility of the full Python language, along with a convenient set of functions to simplify web testing.

SST framework consists of:

  • user actions and assertions (API) in Python
  • test case loader (generates/compiles scripts to unittest cases)
  • console test runner
  • data parameterization/injection
  • selectable output reports
  • selectable browsers
  • headless (xvfb) mode
  • screenshots on errors

Test output can be displayed to the console, saved as an HTML report, or JUnit-compatible XML for compatibility with CI systems.

SST is free open source software (Apache Licensed). SST is primarily being developed on Linux, specifically Ubuntu. It should work fine on other platforms, but any issues (or even better - patches) should be reported on the Launchpad project:

I just uploaded SST 0.1.0 to PyPI:

go ahead, give it a try:

  • `$ [sudo] pip install sst`

documentation and more info:

a sample test script in SST:

from sst.actions import *

assert_title_contains('Ubuntu homepage')

Here is the development progress of SST (shown as a code_swarm visualization) over the past 8 months:

Special thanks to all the SST code committers so far:

  • Danny Tamez
  • Julien Funk
  • Kenneth Koontz
  • Leo Arias
  • Lukasz Czyzykowkski
  • Rick McBride
  • Sidnei da Silva

Extra special thanks to SST's initial creator:

  • Michael Foord

Happy New Year!

-Corey Goldberg