This section details the various ways of installing and compiling Mayavi.
If you already have Mayavi up and running, you can skip this section.
|Windows:||Under Windows the best way to install Mayavi is to install a full Python distribution, such as Enthought Canopy, Pythonxy, or Anaconda. Note that for Pythonxy, you need to check in ‘ETS’ in the installer, when selecting components. If you want to reduce the disk space used by Pythonxy, you can uncheck other components.|
|MacOSX:||The full Python distribution Enthought Canopy (that includes Mayavi) is also available for MacOSX. Unless you really enjoy the intricacies of compilation; this is the best solution to install Mayavi.|
|Ubuntu or Debian:|
|Mayavi is packaged in Debian and Ubuntu. In addition, more up to date packages of Mayavi releases for old versions of Ubuntu are available at https://launchpad.net/~gael-varoquaux/+archive . Experimental Debian packages are also available at http://people.debian.org/~varun/ .|
|RedHat EL3 and EL4:|
|The full Python distribution Enthought Canopy (that includes Mayavi) is also available for RHEL3 and 4.|
If you are not using full, ready-made, scientific Python distribution, you need to satisfy Mayavi’s requirements (for a step-by-step guide on installing all these under windows, see below).
Mayavi requires at the very minimum the following packages:
The following requirements are really optional but strongly recommended, especially if you are new to Mayavi:
One can install the requirements in several ways.
- Windows and MacOSX: even if you want to build from source, a good way to install the requirements is to install one of the distributions indicated above. Note that under Windows, Enthought Canopy comes with a compiler (mingw) and facilitates building Mayavi.
- Linux: Most Linux distributions will have installable binaries available for the some of the above. For example, under Debian or Ubuntu you would need python-vtk, python-wxgtk2.6, python-setuptools, python-numpy, python-configobj.
First make sure you have the prerequisites for Mayavi installed, as indicated in the previous section, i.e. the following packages:
Mayavi is part of the Enthought Tool Suite (ETS). As such, it is distributed as part of ETS and therefore binary packages and source packages of ETS will contain Mayavi. Mayavi releases are almost always made along with an ETS release. You may choose to install all of ETS or just Mayavi alone from a release.
ETS has been organized into several different Python packages. These packages are distributed as Python Eggs. Python eggs are fairly sophisticated and carry information on dependencies with other eggs. As such they are rapidly becoming the standard for distributing Python packages.
The easiest way to install Mayavi with eggs is to use pre-built eggs built for your particular platform and downloaded by easy_install. Alternatively easy_install can build the eggs from the source tarballs. This is also fairly easy to do if you have a proper build environment.
To install eggs, first make sure the essential requirements are installed, and then build and install the eggs like so:
$ easy_install "Mayavi[app]"
This one command will download, build and install all the required ETS related modules that Mayavi needs for the latest ETS release, this means that the Traits dependencies and the Envisage dependencies will be installed automatically.
If you are running a unix system (such as Linux) we advice you not to install the files in the system directories (/usr). An easy way to avoid this is to run:
$ easy_install --prefix "Mayavi[app]"
Automatic downloading of required eggs
If you wish to download all the eggs fetched by easy_install, for instance to propagate to an offline PC, you can use virtualenv to create an empty site-packages, and install to it:
virtualenv --no-site-packages temp cd temp source bin/activate mkdir temp_subdir easy_install -zmaxd temp_subdir "Mayavi[app,nonets]"
If you do not wish to install a ready-made distribution under Windows, these instructions (provided by Guillaume Duclaux) will guide you through the necessary steps to configure a Windows environment in which Mayavi will run.
Install Python 2.5. Add ‘C:\Python25;` to the PATH environment variables.
Install Mingw32, from the Download section of http://www.mingw.org/ , use the MinGW5.1.4 installer. Add ‘C:\MinGW\bin;’ to the PATH environment variables.
Create a ‘c:\documents and settings\USERNAME\pydistutils.cfg’ file(where USERNAME is the login) with the following contents:
Create the new environment variable HOME and set it to the value: ‘c:\docume~1\USERNAME’ (where USERNAME is the login name)
Install Setuptools (0.6c9 binary) from its webpage, and ‘C:Python25Scripts;’ to the PATH environment variables
Install VTK 5.2 (using Dr Charl P. Botha Windows binary http://cpbotha.net/2008/09/23/python-25-enabled-vtk-52-windows-binaries/ )
- Unzip the folder content in ‘C:\Program Files\VTK5.2_cpbotha’
- add ‘C:\Program Files\VTK5.2_cpbotha\bin;’ to the PATH environment variables
- create a new environment variable PYTHONPATH and set it to the value ‘C:\Program Files\VTK5.2_cpbotha\lib\site-packages;’
- If you are running an old version of windows (older than XP) download msvcr80.dll and msvcp80.dll from the www.dll-files.com website and copy them into C:\winnt\system32.
Install Numpy (binary from http://numpy.scipy.org/ )
Installing wxPython (2.8 binary from http://www.wxpython.org/ )
Run in cmd.exe:
easy_install Sphinx EnvisageCore EnvisagePlugins configobj
Finally, run in cmd.exe:
Under Mac OSX Snow Leopard, you may need to build VTK yourself. Here are instructions specific to Snow Leopard (thanks to Darren Dale for providing the instructions):
Download the VTK tarball, unzip it, and make a build directory (vtkbuild) next to the resulting VTK directory
Then cd into vtkbuild and run “cmake ../VTK”. Next, edit CMakeCache.txt (in vtkbuild) and set:
//Build Verdict with shared libraries. BUILD_SHARED_LIBS:BOOL=ON //Build architectures for OSX CMAKE_OSX_ARCHITECTURES:STRING=x86_64 //Minimum OS X version to target for deployment (at runtime); newer // APIs weak linked. Set to empty string for default value. CMAKE_OSX_DEPLOYMENT_TARGET:STRING=10.6 //Wrap VTK classes into the Python language. VTK_WRAP_PYTHON:BOOL=ON //Arguments passed to "python setup.py install ..." during installation. VTK_PYTHON_SETUP_ARGS:STRING=
Run “cmake ../VTK” again.
Run “make -j 2” for a single cpu system. “make -j 9” will compile faster on an 8-core system.
Run “sudo make install”
Edit your ~/.profile and add the following line:
Run “source ~/.profile” or open a new terminal so the DYLD_LIBRARY_PATH environment variable is available.
After that, install Mayavi in the usual way.
If you want to get the latest development version of Mayavi (e.g. for developing Mayavi or contributing to the documentation), we recommend that you check it out from github: Mayavi is hosted on github, with the rest of the Enthought open source packages: the ‘ETS’ (Enthought Tool Suite): https://github.com/enthought
Mayavi depends on several packages that are part of ETS. It is highly likely that the in-development mayavi version may depend on some feature of an as yet unreleased component. Therefore, it is very convenient to get all the relevant ETS projects that Mayavi recursively depends on in one single checkout. For this purpose a script ets.py is available.
Make sure there is no other ETS package installed in your pythonpath:$ python >>> import enthought Traceback (most recent call last): File "<stdin>", line 1, in <module> ImportError: No module named enthought
If you don’t get the ImportError (e.g. importing enthought succeeds), then there is no way to install the git Mayavi version over it (even if you put it first in your PYTHONPATH), because the older (setuptools managed) ETS packages will get picked up too and they will mess up things. This behavior might be surprising if you are new to setuptools.
So for example if you use Ubuntu or Debian, you need to first remove all ETS packages (in Ubuntu 9.04, you need to remove all of these: mayavi2 python-apptools python-enthoughtbase python-envisagecore python-envisageplugins python-traits python-traitsbackendwx python-traitsgui).
Create an empty directory and download in it the ets.py script from https://github.com/enthought/ets/raw/master/ets.py
To get just the sources for mayavi and all its dependencies do this:$ python ets.py clone
This will download from github the source code for the entire ETS.
The ets.py downloads the entire ETS, which is more than you need to build Mayavi. As the extra packages have additional dependencies, they may render the build harder. You can remove safely the following directories:blockcanvas chaco codetools enable graphcanvas scimath
Once the sources are checked out you may either:
Install a development version, to track changes to github easily (recommended):$ python ets.py develop
This will install all the checked out sources by executing a python setup.py develop in each sub directory.
To install of the packages in a different location than the default one, eg ‘~/usr/’, use the following syntax:$ python ets.py develop --prefix ~/usr
make sure that the corresponding site-packages folder is in your PYTHONPATH environment variable (for the above example it would be: ‘~/usr/lib/python2.x/site-packages/’
Alternatively, if you’d like just Mayavi installed via a standard python setup.py install you may do:$ python ets.py develop -f ../dist
You should now have the latest version of Mayavi installed and usable.
The easiest way to test if your installation is OK is to run the mayavi2 application like so:
To get more help on the command try this:
mayavi2 is the mayavi application. On some platforms like win32 you will need to double click on the mayavi2.exe program found in your Python2X\Scripts folder. Make sure this directory is in your path.
Mayavi can be used in a variety of other ways but the mayavi2 application is the easiest to start with.
If you have the source tarball of mayavi or have checked out the sources from the github repository, you can run the examples in mayavi*/examples. There are plenty of example scripts illustrating various features. Tests are available in the mayavi*/tests sub-directory.