Install Instructions

David DeTomaso edited this page Mar 8, 2016 · 6 revisions

Installing the easy way

FastProject is written in Python (2.7 or 3.3+) and has a few dependencies. To simplify the install process, we have created an install file for each platform that will take care of everything.

On Linux or OSX, download the install file and place it in the directory where you want to install FastProject. Navigate to the directory in the terminal and run:

bash FP_OSX_Install.sh

or

bash FP_Linux_Install.sh

depnding on which installer you downloaded.

On Windows, download the install file and place it in the directory where you want to install FastProject. Then, right-click the file and in the popup menu select "Run with PowerShell".

Note: Installation downloads necessary files from the internet. An internet connection is required. Note: The installer will create a new folder, FastProject, in the directory where it is run. So if you want to install in ~/Applications/FastProject for example, place the file in ~/Applications and run it.

Installing FastProject with an existing Python install

The scripts above are provided for convenience. In detail, this is what they do:

  1. Download the Miniconda utility from Continuum
  2. Create a new Python2 install in the $INSTALL_DIR/FastProject directory
  3. Use conda to install dependences (numpy, scripy, scikit-learn, pandas) in the environment
  4. Use pip to install FastProject into the environment
  5. Create a separate Scripts folder in this environment, and provide the option to add the fastproject executable to your path.

In this way, the installer creates an isolated Python environment just for FastProject. This ensures that there are no issues with clashing dependencies. For example, the default Python install on many OSX distributions comes with numpy 1.8 which is not compatible with FastProject and is very difficult to upgrade.

However if you want to install FastProject into an existing Python installation, just execute

pip install FastProject

...and FastProject along with any required dependences (that you don't already have) will be downloaded and installed.

Note: Depending on your platform, pip may be unable to successfully install the 'numpy' dependency of FastProject. For this I recommend using a distribution tool such as 'Conda' which comes with Continuum's Miniconda (available free of charge).

If you are using conda, first install the dependencies by running:

conda install numpy scipy scikit-learn pandas

And then run the "pip install FastProject" command.

Installing FastProject in developer mode

We encourage anyone who would like to play around with the code to fork the FastProject repository on Github.

Then, to install FastProject for convenient developing, first use the "git clone" command to download the code locally. Navigate to the top-level FastProject folder and run:

python setup.py develop

Instead of copying the source files into the "site-packages" directory, the 'develop' argument installs a link in your python's distribution's "site-packages" folder to the cloned repository. This allows you to make changes in the cloned repository and easily test them and commit them.