Which Install is for you?

The full binary install is meant for 32-bit Linux computers.  It installs nothing to system directories, so does not require root access to use.  That version is often very handy for users of shared computational clusters or for users with an older operating system that doesn't package the prerequisites.  Often, it is the installation of the prerequisites, not SincLib itself, which is difficult.

The source code install is meant for folks who have root access to their computers.  It allows them to install the prerequisites on their own to system directories, then only have to download files the SincLib files.

Full Binary Install

If you are running a modern Linux distribution on a 64 or 32-bit computer, you should be able to use our prepackaged binary.  Not only does it include SincLib, but also all the prerequisites.  The binaries are recompiled every night on a development machine.  Let $BASE_DIR represent the base directory which you will be installing too (preferably local disk, but not necessary in a system directory).  To install this, do the following from your installation directory:

cd $BASE_DIR
wget http://t2.unl.edu/store/SincDeps.tar.bz2
tar jxf SincDeps.tar.bz2
cd SincDeps
source setup.sh

Root access is not needed.  Every time you want to use SincLib, you will need to do the following:

cd $BASE_DIR/SincDeps
source setup.sh

The setup.sh script alters your environment so python can find SincLib's dependencies.  Type the following from the $BASE_DIR/SincDeps directory:

> python tests/SincTests.py TestCModule.test_quadratures
SincMethods.SincTests:INFO:  C speedup in Quadrature: 2.540054
.
----------------------------------------------------------------------
Ran 1 test in 0.006s

OK

If everything works, you should see the word OK written at the end.  Due to the complexity of the source install, we ask you to please let us know if the binaries do not work.

Source Install

The reason that the binary option was introduced is that many users get stuck trying to install the prerequisites, which may not have been packaged with their operating system. 

Installing prerequisites on Fedora Core 7 or later

Fedora Core 7 or later are the best operating systems to try SincLib on because they package the prerequisites.  To install them, do the following:

yum install numpy scipy python-matplotlib

If the svn binary is not present on the system, additionally do this:

yum install subversion

Source code download

It is often handy to keep all your source code in one place.  I will assume that you are in a directory called ~/projects/.
First, download the following packages from the local repository:

svn co svn://t2.unl.edu/brian/SincMethods/trunk SincMethods

You will only need to do this when you originally install SincLib.  To update an already-installed directory to the latest source code, do the following:

svn up

Finally, each time you log in to a new shell and would like to run SincLib, you must cd to the SincMethods directory and do the following:

source setup.sh

Alternately, some users like to put the following line in their .bashrc file, which eliminates the need to source the setup.sh script each time:

export PYTHONPATH=/path/to/SincMethods/src:$PYTHONPATH

You will need to log out / log in for that to take effect.

Compiling C Modules

SincLib relies on several C modules underneath.  These must be built on your computer, but you do not need root permissions to do this.  Execute this command:

python setup.py build

from the SincMethods directory.