Thanks for the instructions. Following them I was able to set up my environment, run your new code with a few Pegasus dax files, and create some graphml files.

If you want to save some time replace the first line above with

conda create --name myenv networkx six

which install only necessary packages to test dax2graphml.py (I used your sim.dax from DM-9092).

I updated the source code getting rid of this pesky unicode character. It should work fine on lsst-dev now with either Python 2 or 3.

To make xmltodict available on lsst-dev you need to create your own environment first. Though it may be an overkill, the easiest way to do it is to clone existing one called root. After you clone it, say as myenv, you can switch to it and install packages at will. Here are the commands to perform each step:

conda create --name myenv --clone root

source activate myenv

pip install xmltodict

It imports fine for me with Python 3 but I also can't see that character in the source.

Two quick comments/questions for now:

1. I encountered this error when trying to run it:

 File "daxgen/utils/dax2graphml.py", line 132

SyntaxError: Non-ASCII character '\xc2' in file daxgen/utils/dax2graphml.py on line 133, but no encoding declared; see http://python.org/dev/peps/pep-0263/ for details

Can you fix that please?

2. I use the Python that comes with the shared stack, and it does not have xmltodict. I cannot install new package to the shared stack. Which python do you use on lsst-dev?

I added a simple utility to my DAX generator which allows me to convert (your) existing Pegasus plans to GraphML e.g. for testing purposes.

I would appreciate if you could take a look at the code.