Skip to content

Latest commit

 

History

History
108 lines (80 loc) · 3.79 KB

preworkshop.md

File metadata and controls

108 lines (80 loc) · 3.79 KB

Pre-workshop Instructions

We would like the workshop to be an interactive experience, were you get to try everything yourself, rather than just watch someone else doing it. For this to happen, you will need to do some homework prior to the workshop itself.

If you have trouble with the instructions below, there will be some volunteers available before and during the workshop. Plan on coming up to half an hour early to get help with your issues.

  1. Bring a laptop

To take full advantage of the workshop, you will want to have a laptop with you. Many of the things we will be trying out are (much) harder to get to work under Windows, so if you have the option you want to bring a Linux or OS X computer.

  1. Install Git

You will need to have Git working on your system. Unless you are using Windows, you probably already have it on your system. You can find detailed instructions for any OS on Git's website.

  1. Join GitHub

While there are other options out there, many open source projects in Python's data science ecosystem are hosted on GitHub. If you don't already, you will want to have an account set up with them. This is completely free, but you will need to provide your e-mail address when signing up. Follow the instructions on GitHub's website.

  1. Install Python

You will need a Python distribution installed on your system. Actually, if you plan on doing any kind of serious development, you will likely end up having a few of them. There are several tools that help manage this, each with its advantages and disadvantages.

For the purpose of this workshop, we will use Continuum's Anaconda distribution, and their conda package manager. To use this recommended setup, start by installing Miniconda, following the instructions for your OS on conda's documentation.

After installing and updating conda itself per the above instructions, create an environment for NumPy development running from the command line:

conda create -n numpydev35 python=3.5 nose cython

This will create an environment, called numpydev35, with the latest Python version (3.5), and the dependencies required to build and test the NumPy library.

If you are a Python 2.x kind of person, requesting python=2.7 (and changing the environment name accordingly) should do the trick for you.

After creating the environment, you can activate it by running from the command line

source activate numpydev35

on Linus or OS X, or

activate numpydev35

under Windows.

  1. Install a C compiler

Building NumPy requires a C compiler, so you want to make sure one is properly installed.

  • If you are on a Linux system, chances are gcc is already installed, type gcc --version on the command line to confirm.
  • On OS X, you will need the xcode toolset. To check if it is installed, run the same gcc --version from the command line. On Yosemite, if xcode is not available, this will display a pop-up window requesting permission to install it. For earlier versions, manually download and install it.
  • If you are a Windows user, you are about to enter a world of pain. Your best bet is installing MinGW, although this is far from a perfect solution. If you go this route, do not install the latest Python version, settle with 3.4, as MinGW doesn't support 3.5 yet. You will also need to make sure that your conda installation is a 32 bit one, as the 64 bit toolchain is broken.