Want to install code with Python?

pip, virtualenv, and virtualenvwrapper are all great tools, but their out-of-the-box configuration, especially for new users, is basically broken. All the useful, advanced features that experienced python developers use are hidden behind a bunch of command-line options you need to search around in the documentation to find, and even then, some of them have to match each other in ways which are not immediately clear.

In the future, pip should be easier to use, but right now some of these features are bogged down in long discussions.

So welcome to the web site of the PIP OF THE FUTURE!

The following shell script and Mac application will set up your command-line environment to run pip in such a way that everything works by default.

What does this do and how do I use it?

You need a Mac OS X or Linux system with Python already installed.

Download the shell script or application and run it. (The application is currently unsigned, so on a mac with default settings, you may need to right-click and go to 'open' rather than double click; sorry about that.)

Once it has run and installed pip, you should be able to just pip install whatever you like. This will, by default, install code into your home directory.

However, if you wish to have a clean python environment for each project you're working on, you can also create virtualenvs with the mkvirtualenv command, as usual for virtualenvwrapper. If you are working on a virtualenv, pip install will install into the active virtualenv and not your home directory.

If you have to install things repeatedly into virtualenvs, pip wheel foo before pip install foo will give you extra-fast installs.

Who did this and why?

This project is by Glyph.

I created this as a proof of concept of several things:

• It's really annoying for teachers of Python workshops to set up people's computers. This was designed primarily to help them by giving them a one-click way to get started with pip.
• When teachers of Python do set up students' computers, they tend to do it completely differently, and often in ways which break other Python applications. This was designed to provide a consistent, non-invasive way to get things set up for development.
• By default, pip just emits tracebacks when you attempt to run it without privileges. Running pip is (A) sometimes not possible and (B) always a bad idea because it might screw up your operating system. The defaults in the situation we should be encouraging users to set up should be well-behaved.
• Many tutorials recommend installing external versions of Python onto Mac OS X just to get it to work. However, the system comes with a perfectly usable Python, and it's a lot quicker to set up a new user. The one sticking point is the presence of an antediluvian setuptools in the system. This automation includes an extremely subtle hack to work around this which has the happy accident of working around similarly broken linux distributions.
• Pip's default configuration disables many of its more interesting features (wheel-building, download caching, wheel caching, installing files from the wheel cache).

But now I need a C compiler for this Python package I want to use!

If you're on Debian or Ubuntu, sudo apt-get install build-essential python-dev.

If you're on the Mac, hopefully you're on Mavericks. If so, type xcode-select --install in a Terminal window.

If you're on Fedora, sudo yum install python-devel.

But what about Windows?

I feel your pain, really, I do. But Windows is an entirely different, unrelated set of problems. Particularly, getting a C compiler that matches your Python version is nearly impossible, and outside the scope of issues this tool intends to address.

What is the license?

The shell script itself is available under the MIT license.

The Mac application is available under the GPLv2, as per the terms of the license of Platypus, the tool used to create it.