Installation

Make things as simple as possible, but not any simpler.

—Albert Einstein

The Python Distutils are used to build and install PyTables, so it is fairly simple to get the application up and running. If you want to install the package from sources you can go on reading to the next section.

However, if you want to go straight to binaries that ‘just work’ for the main platforms (Linux, Mac OSX and Windows), you might want to use the excellent Anaconda or Canopy distributions. PyTables usually distributes its own Windows binaries too; go Binary installation (Windows) for instructions. Finally Christoph Gohlke also maintains an excellent suite of a variety of binary packages for Windows at his site.

Installation from source

These instructions are for both Unix/MacOS X and Windows systems. If you are using Windows, it is assumed that you have a recent version of MS Visual C++ compiler installed. A GCC compiler is assumed for Unix, but other compilers should work as well.

Extensions in PyTables have been developed in Cython (see [CYTHON]) and the C language. You can rebuild everything from scratch if you have Cython installed, but this is not necessary, as the Cython compiled source is included in the source distribution.

To compile PyTables you will need a recent version of Python, the HDF5 (C flavor) library from http://www.hdfgroup.org, and the NumPy (see [NUMPY]) and Numexpr (see [NUMEXPR]) packages.

Prerequisites

First, make sure that you have

  • Python >= 2.6 including Python 3.x
  • HDF5 >= 1.8.4 (>=1.8.7 is strongly recommended),
  • NumPy >= 1.4.1,
  • Numexpr >= 2.0 and
  • Cython >= 0.13
  • argparse (only Python 2.6, it is used by the pt2to3 utility)

installed (for testing purposes, we are using HDF5 1.8.12, NumPy 1.8.0 and Numexpr 2.2.2 currently). If you don’t, fetch and install them before proceeding.

Note

HDF5 versions < 1.8.7 are supported with some limitations. It is not possible to open the same file multiple times (simultaneously), even in read-only mode.

Note

Currently PyTables does not use setuptools by default so do not expect that the setup.py script automatically install all packages PyTables depends on.

Compile and install these packages (but see Windows prerequisites for instructions on how to install pre-compiled binaries if you are not willing to compile the prerequisites on Windows systems).

For compression (and possibly improved performance), you will need to install the Zlib (see [ZLIB]), which is also required by HDF5 as well. You may also optionally install the excellent LZO compression library (see [LZO] and Compression issues). The high-performance bzip2 compression library can also be used with PyTables (see [BZIP2]).

The Blosc (see [BLOSC]) compression library is embedded in PyTables, so this will be used in case it is not found in the system. So, in case the installer warns about not finding it, do not worry too much ;)

Unix

setup.py will detect HDF5, LZO, or bzip2 libraries and include files under /usr or /usr/local; this will cover most manual installations as well as installations from packages. If setup.py can not find libhdf5, libhdf5 (or liblzo, or libbz2 that you may wish to use) or if you have several versions of a library installed and want to use a particular one, then you can set the path to the resource in the environment, by setting the values of the HDF5_DIR, LZO_DIR, BZIP2_DIR or BLOSC_DIR environment variables to the path to the particular resource. You may also specify the locations of the resource root directories on the setup.py command line. For example:

--hdf5=/stuff/hdf5-1.8.12
--lzo=/stuff/lzo-2.02
--bzip2=/stuff/bzip2-1.0.5
--blosc=/stuff/blosc-1.3.2

If your HDF5 library was built as a shared library not in the runtime load path, then you can specify the additional linker flags needed to find the shared library on the command line as well. For example:

--lflags="-Xlinker -rpath -Xlinker /stuff/hdf5-1.8.12/lib"

You may also want to try setting the LD_LIBRARY_PATH environment variable to point to the directory where the shared libraries can be found. Check your compiler and linker documentation as well as the Python Distutils documentation for the correct syntax or environment variable names. It is also possible to link with specific libraries by setting the LIBS environment variable:

LIBS="hdf5-1.8.12 nsl"

Finally, you can give additional flags to your compiler by passing them to the --cflags flag:

--cflags="-w -O3 -msse2"

In the above case, a gcc compiler is used and you instructed it to suppress all the warnings and set the level 3 of optimization. Finally, if you are running Linux in 32-bit mode, and you know that your CPU has support for SSE2 vector instructions, you may want to pass the -msse2 flag that will accelerate Blosc operation.

Hint

some GNU/Linux distributions provide a packaged version of the HDF5 libraries with MPI support. In this case you may need to specify the path of the MPH headers as additional include directory.

On Ubuntu 12.04 the following command has been reported to work:

$ C_INCLUDE_PATH=/usr/lib/openmpi/include pip install --upgrade tables

Windows

You can get ready-to-use Windows binaries and other development files for most of the following libraries from the GnuWin32 project (see [GNUWIN32]). In case you cannot find the LZO binaries in the GnuWin32 repository, you can find them at http://sourceforge.net/projects/pytables/files/lzo-win. Once you have installed the prerequisites, setup.py needs to know where the necessary library stub (.lib) and header (.h) files are installed. You can set the path to the include and dll directories for the HDF5 (mandatory) and LZO, BZIP2, BLOSC (optional) libraries in the environment, by setting the values of the HDF5_DIR, LZO_DIR, BZIP2_DIR or BLOSC_DIR environment variables to the path to the particular resource. For example:

set HDF5_DIR=c:\\stuff\\hdf5-1.8.5-32bit-VS2008-IVF101\\release
set LZO_DIR=c:\\Program Files (x86)\\GnuWin32
set BZIP2_DIR=c:\\Program Files (x86)\\GnuWin32
set BLOSC_DIR=c:\\Program Files (x86)\\Blosc

You may also specify the locations of the resource root directories on the setup.py command line. For example:

--hdf5=c:\\stuff\\hdf5-1.8.5-32bit-VS2008-IVF101\\release
--lzo=c:\\Program Files (x86)\\GnuWin32
--bzip2=c:\\Program Files (x86)\\GnuWin32
--blosc=c:\\Program Files (x86)\\Blosc

Development version (Unix)

Installation of the development version is very similar to installation from a source package (described above). There are two main differences:

  1. sources have to be downloaded from the PyTables source repository hosted on GitHub. Git (see [GIT]) is used as VCS. The following command create a local copy of latest development version sources:

    $ git clone https://github.com/PyTables/PyTables.git
  2. sources in the git repository do not include pre-built documentation and pre-generated C code of Cython extension modules. To be able to generate them, both Cython (see [CYTHON]) and sphinx >= 1.0.7 (see [SPHINX]) are mandatory prerequisites.

PyTables package installation

Once you have installed the HDF5 library and the NumPy and Numexpr packages, you can proceed with the PyTables package itself.

  1. Run this command from the main PyTables distribution directory, including any extra command line arguments as discussed above:

    $ python setup.py build
  2. To run the test suite, execute any of these commands.

    Unix

    In the sh shell and its variants:

    $ cd build/lib.linux-x86_64-3.3
    $ env PYTHONPATH=. python tables/tests/test_all.py

    or, if you prefer:

    $ cd build/lib.linux-x86_64-3.3
    $ env PYTHONPATH=. python -c "import tables; tables.test()"

    Note

    the syntax used above overrides original contents of the PYTHONPATH environment variable. If this is not the desired behaviour and the user just wants to add some path before existing ones, then the safest syntax to use is the following:

    $ env PYTHONPATH=.${PYTHONPATH:+:$PYTHONPATH} python tables/tests/test_all.py

    Please refer to your sh documentation for details.

    Windows

    Open the command prompt (cmd.exe or command.com) and type:

    > cd build\\lib.linux-x86_64-2.7
    > set PYTHONPATH=.;%PYTHONPATH%
    > python tables\\tests\\test_all.py

    or:

    > cd build\\lib.linux-x86_64-2.7
    > set PYTHONPATH=.;%PYTHONPATH%
    > python -c "import tables; tables.test()"

    Both commands do the same thing, but the latter still works on an already installed PyTables (so, there is no need to set the PYTHONPATH variable for this case). However, before installation, the former is recommended because it is more flexible, as you can see below. If you would like to see verbose output from the tests simply add the -v flag and/or the word verbose to the first of the command lines above. You can also run only the tests in a particular test module. For example, to execute just the test_types test suite, you only have to specify it:

    # change to backslashes for win
    $ python tables/tests/test_types.py -v

    You have other options to pass to the test_all.py driver:

    # change to backslashes for win
    $ python tables/tests/test_all.py --heavy

    The command above runs every test in the test unit. Beware, it can take a lot of time, CPU and memory resources to complete:

    # change to backslashes for win
    $ python tables/tests/test_all.py --print-versions

    The command above shows the versions for all the packages that PyTables relies on. Please be sure to include this when reporting bugs:

    # only under Linux 2.6.x
    $ python tables/tests/test_all.py --show-memory

    The command above prints out the evolution of the memory consumption after each test module completion. It’s useful for locating memory leaks in PyTables (or packages behind it). Only valid for Linux 2.6.x kernels. And last, but not least, in case a test fails, please run the failing test module again and enable the verbose output:

    $ python tables/tests/test_<module>.py -v verbose

    and, very important, obtain your PyTables version information by using the --print-versions flag (see above) and send back both outputs to developers so that we may continue improving PyTables. If you run into problems because Python can not load the HDF5 library or other shared libraries.

    Unix

    Try setting the LD_LIBRARY_PATH or equivalent environment variable to point to the directory where the missing libraries can be found.

    Windows

    Put the DLL libraries (hdf5dll.dll and, optionally, lzo1.dll, bzip2.dll or blosc.dll) in a directory listed in your PATH environment variable. The setup.py installation program will print out a warning to that effect if the libraries can not be found.

  3. To install the entire PyTables Python package, change back to the root distribution directory and run the following command (make sure you have sufficient permissions to write to the directories where the PyTables files will be installed):

    $ python setup.py install

    Of course, you will need super-user privileges if you want to install PyTables on a system-protected area. You can select, though, a different place to install the package using the --prefix flag:

    $ python setup.py install --prefix="/home/myuser/mystuff"

    Have in mind, however, that if you use the --prefix flag to install in a non-standard place, you should properly setup your PYTHONPATH environment variable, so that the Python interpreter would be able to find your new PyTables installation. You have more installation options available in the Distutils package. Issue a:

    $ python setup.py install --help

    for more information on that subject.

That’s it! Now you can skip to the next chapter to learn how to use PyTables.

Installation with pip

Many users find it useful to use the pip program (or similar ones) to install python packages.

As explained in previous sections the user should in any case ensure that all dependencies listed in the Prerequisites section are correctly installed.

The simplest way to install PyTables using pip is the following:

$ pip install tables

The following example shows how to install the latest stable version of PyTables in the user folder when a older version of the package is already installed at system level:

$ pip install --user --upgrade tables

The –user option tels to the pip tool to install the package in the user folder ($HOME/.local on GNU/Linux and Unix systems), while the –upgrade option forces the installation of the latest version even if an older version of the package is already installed.

The pip tool can also be used to install packages from a source tar-ball:

$ pip install tables-3.0.0.tar.gz

To install the development version of PyTables from the develop branch of the main git [GIT] repository the command is the following:

$ pip install git+https://github.com/PyTables/PyTables.git@develop#egg=tables

A similar command can be used to install a specific tagged fersion:

$ pip install git+https://github.com/PyTables/PyTables.git@v.2.4.0#egg=tables

Finally, PyTables developers provide a requirements.txt file that can be used by pip to install the PyTables dependencies:

$ wget https://raw.github.com/PyTables/PyTables/develop/requirements.txt
$ pip install -r requirements.txt

Of course the requirements.txt file can be used to install only python packages. Other dependencies like the HDF5 library of compression libraries have to be installed by the user.

Binary installation (Windows)

This section is intended for installing precompiled binaries on Windows platforms. You may also find it useful for instructions on how to install binary prerequisites even if you want to compile PyTables itself on Windows.

Windows prerequisites

First, make sure that you have Python 2.6, NumPy 1.4.1 and Numexpr 2.0 or higher installed (PyTables binaries have been built using NumPy 1.7 and Numexpr 2.1). The binaries already include DLLs for HDF5 (1.8.4, 1.8.9), zlib1 (1.2.3), szlib (2.0, uncompression support only) and bzip2 (1.0.5) for Windows (2.8.0). The LZO DLL can’t be included because of license issues (but read below for directives to install it if you want so).

To enable compression with the optional LZO library (see the Compression issues for hints about how it may be used to improve performance), fetch and install the LZO from http://sourceforge.net/projects/pytables/files/lzo-win (choose v1.x for Windows 32-bit and v2.x for Windows 64-bit). Normally, you will only need to fetch that package and copy the included lzo1.dll/lzo2.dll file in a directory in the PATH environment variable (for example C:\WINDOWS\SYSTEM) or python_installation_path\Lib\site-packages\tables (the last directory may not exist yet, so if you want to install the DLL there, you should do so after installing the PyTables package), so that it can be found by the PyTables extensions.

Please note that PyTables has internal machinery for dealing with uninstalled optional compression libraries, so, you don’t need to install the LZO dynamic library if you don’t want to.

PyTables package installation

Download the tables-<version>.win32-py<version>.exe file and execute it.

Binary packahes can be found e.g. at the Unofficial Windows Binaries for Python Extension Packages page.

You can (and you should) test your installation by running the next commands:

>>> import tables
>>> tables.test()

on your favorite python shell. If all the tests pass (possibly with a few warnings, related to the potential unavailability of LZO lib) you already have a working, well-tested copy of PyTables installed! If any test fails, please copy the output of the error messages as well as the output of:

>>> tables.print_versions()

and mail them to the developers so that the problem can be fixed in future releases.

You can proceed now to the next chapter to see how to use PyTables.