Added link to CVXOPT in get-build-install documentation + progress bar to rips-pairwise-cohomology
.. _download:
Get, Build, Install
===================
The most up to date code is available from
`my Mercurial repository`_.
If you have Mercurial_, the easiest way to obtain the code is by cloning it:
.. parsed-literal::
hg clone |dionysus-url|
cd Dionysus
hg up tip
If you don't have time or desire to deal with Mercurial, you can download the
`tarball of the entire repository`_. The advantage of using Mercurial is that it
makes it very easy to keep up with the updates that are periodically committed
to the repository::
hg pull -u
.. |dionysus-url| replace:: http://hg.mrzv.org/Dionysus/
.. _Mercurial: http://www.selenic.com/mercurial/
.. _`tarball of the entire repository`: http://hg.mrzv.org/Dionysus/archive/tip.tar.gz
.. _`my Mercurial repository`: http://hg.mrzv.org/Dionysus/
Dependencies
------------
Dionysus requires the following software:
:CMake_: for building (version :math:`\geq` 2.6)
:Boost_: C++ utilities (version :math:`\geq` 1.36; including Boost.Python used to create
Python bindings)
.. warning::
There also seems to be a dependence on the version of GCC, although I don't
entirely understand it. GCC 4.3 and above definitely work, but some versions
below that seem not to.
One particular catch is that the default compiler on many current Mac OS X
is GCC 4.0 that has a well-known bug making Dionysus unusable. Fortunately
the problem is easy to solve by using GCC 4.2 that is often available on a
Mac under the name ``gcc-4.2``.
One can check the compiler version with ``g++ --version`` command.
Optional dependencies:
:CGAL_: for alpha shapes (version :math:`\geq` 3.4)
:CVXOPT_: for :ref:`circle-valued parametrization <cohomology-parametrization>` using LSQR
:rlog_: used for logging only (not needed by default)
.. :dsrpdb_: for reading PDB files
:SYNAPS_: for solving polynomials (for kinetic kernel), which in
turn requires GMP_
.. _CMake: http://www.cmake.org
.. _Boost: http://www.boost.org
.. _CGAL: http://www.cgal.org
.. _CVXOPT: http://abel.ee.ucla.edu/cvxopt/
.. _rlog: http://www.arg0.net/rlog
.. _dsrpdb: http://www.salilab.org/~drussel/pdb/
.. _SYNAPS: http://www-sop.inria.fr/galaad/synaps/
.. _GMP: http://gmplib.org/
Building
--------
To build the examples as well as the :ref:`Python bindings <python-bindings>`,
create a directory ``build``. Inside that directory run ``cmake`` and ``make``::
mkdir build
cd build
cmake ..
make
.. tip::
To use GCC 4.2 on a Mac one can try ``CXX=g++-4.2 cmake ..`` instead of
``cmake ..``.
Instead of ``cmake``, one can run ``ccmake`` for a curses interface. The
following configuration options are available. One can set them either through
the curses interface or by passing a flag of the form ``-Doptimize:bool=on`` to
``cmake``.
:debug: Turns on debugging compilation
:optimize: Turns on compiler optimizations (`on` by default)
:logging: Turns on logging facilities
:counters: Turns on various built-in counters
Depending on the combination of debugging and optimization, a particular
``CMAKE_CXX_FLAGS*`` is chosen.
.. tip:: The default settings work fine unless you want to dive into the
library's internals with logging or study the performance of various
algorithms with counters.
.. todo:: Write sections on logging and counters.
Some parts of Dionysus understand the ``DEBUG_CONTAINERS`` definition which can
be appended to ``CMAKE_CXX_FLAGS``. If set, the library will use GCC STL's
debugging containers (from the ``std::__debug`` namespace defined in ``debug/*``
header files). These containers return safe iterators (the kind that check
whether they are singular when compared, or additionally whether they are out of
bounds when dereferenced).
.. todo:: ``ZIGZAG_CONSISTENCY`` definition
Install
-------
At the moment there are no installation procedures. To run the Python code you
need to have ``.../build/bindings/python`` somewhere in your ``PYTHONPATH``.
I.e. add::
export PYTHONPATH=.../build/bindings/python
to your ``~/.bashrc`` (assuming you are using Bash_). Alternatively, run the
python examples from within ``.../build/bindings/python``::
python .../Dionysus/examples/triangle/triangle.py
The C++ examples can be run from anywhere. The C++ library consists only of
header files (no library actually needs to be built), so to compile against it,
it suffices to add ``-I .../Dionysus/include`` to your ``g++`` flags::
g++ your-code.cpp -o your-code -I .../Dionysus/include
Proper installation procedures (with ``make install``) will be added in the
future.
.. _Bash: http://www.gnu.org/software/bash/