Carneades sample code

This is a pedagogically-oriented attempt to implement aspects of the Carneades Argument Evaluation framework. It closely follows the Haskell implementation in the CarneadesDSL package.

Quick Link:

Documentation for setting up:

1. on DICE machine
2. on your own machine The main difference between them is the creation of python virtual environment.

Requirements

• Python3.4
• igraph
• pycairo (for igraph)
• Virtualenv (Optional)
• Sphinx (docs only)
• Basicstrap theme for sphinx (docs only)

Setting it up on DICE

###Creating python virtual environment Create a python3.4 virtual environment:

S1987654: pyvenv ailp_env # create an python3.4 environment in the folder 'ailp_env'

Activate the python environment (you have to activate the environment everytime before you run Carneades):

S1987654: source ailp_env/bin/activate
# Once the environment is activate, you should see the name of the environment beside it:
(ailp_env) S1987654:

Installing libraries:

Similar to installing on your own computer, you need to install the libraries required. However, there is no need to set up the virutal environment again. I.e. follow the instructions from installing sphinx to installing pycairo

Installation on your own computer

Install python3.4 (via apt, homebrew or your favourite package_system)

Note for Linux users (especially Ubuntu 12.04 users): follow this link to have more info about installing the python distro.

Create a virtualenv:

$ virtualenv -p python3.4 envname
$ source /envname/bin/activate

Install sphinx and basicstrap

$ pip install sphinx
$ pip install sphinxjp.themes.basicstrap

Install igraph

$ pip install python-igraph

This should also install the C bindings. If that doesn't happen because of some error, it's likely to be one of the following problems:

• The header files for the python distribution are not installed. Look up build-essential, and python-dev or python3.4-dev

# in a new terminal:
$ sudo apt-get install python-dev python3.4-dev
$ sudo apt-get install build-essential

• To install the C bindings:

$ sudo apt-get install libxml2-dev
# download the igraph c bindings and extract.
$ wget http://igraph.org/nightly/get/c/igraph-0.7.1.tar.gz
$ tar xvf igraph-0.7.1.tar.gz
$ cd igraph-0.7.1

# Installation
$ ./configure
$ make
$ make install
$ sudo ldconfig # this does some magic!

Install pycairo

On Ubuntu, check this package.

If you are using a virtualenv or you don't use Ubuntu, do the following:

• Download and extract the pycairo package in the virtualenv

$ wget https://www.cairographics.org/releases/pycairo-1.10.0.tar.bz2
$ bunzip2 pycairo-1.10.0.tar.bz2
$ tar xvf pycairo-1.10.0.tar
$ cd pycairo-1.10.0/
$ ./waf configure # This will fail, but need to generate the hidden .waf folder first for below

• edit a file in the hidden .waf folder that the extraction has created (e.g. ./.waf-1.6.4-e3c1e08604b18a10567cfcd2d02eb6e6. To do this, go into the folder and edit ./.waf-1.6.4-somenumbers/waflib/Tools/python.py and replace the following line (see -/+ in codeblock below) to call the python3.4-config directly.

  --- waflib/Tools/python.py.old  2014-08-01 14:36:23.750613874 +0000
    +++ waflib/Tools/python.py      2014-08-01 14:36:38.359627761 +0000
    @@ -169,7 +169,7 @@
                    conf.find_program('python-config-%s'%num,var='PYTHON_CONFIG',mandatory=False)
            includes=[]
            if conf.env.PYTHON_CONFIG:
    -               for incstr in conf.cmd_and_log(conf.env.PYTHON+[conf.env.PYTHON_CONFIG,'--includes']).strip().split():
    +               for incstr in conf.cmd_and_log([conf.env.PYTHON_CONFIG,'--includes']).strip().split():
                            if(incstr.startswith('-I')or incstr.startswith('/I')):
                                    incstr=incstr[2:]
                            if incstr not in includes:

• compile and install the package

$ ./waf configure --prefix=$VIRTUAL_ENV
$ ./waf build
$ ./waf install

• If you run into 'cairo >= 1.10.0' error, you will need to install cairo and pixman:

Download and Install pixman

# Open a new terminal and run these commands:
$ wget https://www.cairographics.org/releases/pixman-0.34.0.tar.gz
$ tar xvf pixman-0.34.0.tar.gz
$ cd pixman-0.34.0

# configure make install:
$ ./configure
$ make
$ make install

Download and Install cairo

# continuing from above:
$ cd ..
$ wget https://www.cairographics.org/releases/cairo-1.14.6.tar.xz
$ tar xvf cairo-1.14.6.tar.xz
$ cd cairo-1.14.6

# configure make install:
$ ./configure
$ make
$ make install

You should now have both cairo and pixman installed in your machine. Return to installation for pycairolink.

Clone repository

$ git clone https://github.com/ewan-klein/carneades.git
$ cd carneades

Generate the documentation

$ cd doc
$ make html

Test the software

$ cd ../src/carneades # relative root is carneades/
$ python -i caes.py

Install python3.4

(Linux users)

The following is to install python3.4 on Ubuntu 12.04 LTS. Other distributions should do more or less the same. If you have a newer installation of Ubuntu or derivatives, you can probably skip some passages.

WARNING: Although unlikely, this might break your python2 installation, so do it at your own risk or set up a VM

Unless you want to build from source (In which case, kudos!), add this PPA to your apt source list:

$ add-apt-repository ppa:fkrull/deadsnakes
$ apt-get update

then run:

$ apt-get install python3.4 python3.4-dev

Now follow the rest of the installation.

Work around for pycairo and igraph

For output where the graphs are not shown, here is a workaround to produce the graph. Thanks to the write_to_graphviz method, a .dot file will be produced whenever you run caes.py. This .dot file can then be visualised using a Graphiviz viewer, such as:

• http://www.webgraphviz.com/
• http://sandbox.kidstrythisathome.com/erdos/
• http://dreampuf.github.io/GraphvizOnline/

The .dot file is by default graph.dot. It will be good to rename each caes example with a different filename to prevent it being overwritten in future iterations of caes.py.

# In caes.py, e.g. in def arg_demo(), change the following line to define the filename
>> argset.draw()
>> argset.write_to_graphviz(<filename>.dot)

The <filename>.dot file will be found adjacent to the caes.py file. (defaul in $PATH_TO_CARNEADES/carneades/src/carneades).