OK, I went ahead and merged this. Thanks for the comments, guys!

I like the direction of this, and I think it's best if you agree on the details between yourselves. In other words, looks good to me.

"Ah, now I understand your comment from msg9034. I don't think we usually maintain the old experiment scripts."

I just saw that you changed the "old experiment script" from issue899 by adjusting the print statement and thought your intention is to update those scripts that are mentioned as a reference for some lab feature in the README file (issue 899 is one of them, as is issue 735) because those are the scripts people base their experiments on. Never mind if this was not your intention. 

If it is likely that a lab script doesn't work anymore, I don't think it should be mentioned as a reference in the README, though.

Ah, now I understand your comment from msg9034. I don't think we usually maintain the old experiment scripts.

Regarding msg9036: these commands are for creating the virtual environment. So the second command only works if "python" points to a python3 interpreter, for example, if you're already in a virtual environment. Python 2 doesn't have the venv module.

So this

python3 -m venv --prompt=issue899 venv

and this

python -m venv --prompt=issue899 venv

does exactly the same thing?

One of the benefits of virtual environments is that "python" points to the correct interpreter anyway.

You could think about changing to python 3 in all experiment scripts that are mentioned in the README (like you did in issue899/common_setup.py). For instance, custom_parser.py in issue 735 still has a python 2 print in it (I didn't look any further).

Looks fine for me.

I left a comment.

Pull request: https://bitbucket.org/jendrikseipp/downward/pull-requests/145

Python virtual environments allow us to separate the Lab (and Matplotlib) versions for different experiments, which makes it easier to reproduce experiments later. I'd like to include instructions for this in the experiments/README file.