Installing PyOLS is as simple as:

$ sudo python setup.py install

PyOLS can now be started by running:

$ pyols -c env_dir
$ pyols !$

It is worth examining the output of pyols --help to see the options that are available (especially the --wrapper option).

After PyOLS is installed, init.d/README.txt will help you start PyOLS when your system starts.

If the SCGI wrapper is going to be used a webserver needs to be setup to act as the frontend.

While many web servers support the SCGI proticol, I will only describe the setup procedure for Apache here. Instructions for other servers, such as lighttpd, can be easily found on Google.

1. Make sure the SCGI module is installed (for example, on Debian, run: apt-get isntall libapache2-mod-scgi).
2. Enable the SCGI module. With Apache 2 on Debian, the command is a2enmod scgi. On other distributions, a line similar to LoadModule scgi_module mod_scgi.so will need to be added to apache.conf (or httpd.conf, or modules.conf, depending on your distribution's naming standard).
3. Add the line SCGIMount /$PATH $HOST:$PORT to apache.conf (or httpd.conf, depending on your distribution's naming scheme), where $PATH is replaced by the path which PyOLS will be accessable from (eg, /pyols), $HOST is the host PyOLS will be running on (probably localhost) and $PORT is the port PyOLS will be running on (probably 4000).

When developing with PyOLS (that is, writing an application which calls PyOLS), it will be useful to have a copy of the PyOLS API documentation close at hand. This can be found in the doc/api subdirectory of the tarball. The most relevant file will be be doc/api/pyols.api.OntologyTool-class.html, which contains all the functions available over RPC.

To get started working with PyOLS, setup a "development" installation:

$ python setup.py develop

This will do two things: install a pyols package which will link to the current directory (so there is no need to fuss with symlinks or re-run setup.py after each change), and it will install the nose and pyparsing packages (used for testing)

Running the PyOLS unit tests is simple... So you'd better do it! Simply run nosetests in whatever directory you want tested. For example:

~/pyOLS/pyols/$ nosetests
...........................................
----------------------------------------------------------------------
Ran 43 tests in 6.059s

OK

Many files also contain doctests, and at the bottom of these files there will be the lines:

from pyols.tests import run_doctests
run_doctests()

As the name suggests, this will run doctets on the module (but, if you care to look at the implementation of run_doctests, only if the module is being run directly from the command line).

So, after editing a function with doctests, always update the doctests with the new functionality then execute the file (python foo.py) to verify that both your changes and doctests are correct.

The Namespace.copy_to method makes some SQLite-specific assumptions... Check it if you are migrating to a new database.