Please do not change my matplotlib backend

[DavidMStraub]

Thanks for this package!

I only have one problem with it: at the top of network.py, you are setting

matplotlib.use('Agg')

According to the comment, this is to fix Travis. I guess you are switching to a backend that doesn't need an X server. However, this can be very disruptive to anybody importing your library, e.g. in a Jupyter notebook, where plots stop appearing:

UserWarning: Matplotlib is currently using agg, which is a non-GUI backend,
so cannot show the figure

It took me quite a while to find out that this line from pandana was the culprit. Please remove it.

If it's just about testing on Travis, you can just set the display, it works well, I do it too.

Thanks!

[smmaurer]

Hi @DavidMStraub, thanks for pointing this out! I didn't know it was in there. Definitely agree that we should remove it.

To dos:

• [x] Confirm there won't be problematic side effects from removing the matplotlib setting
• [x] Remove it, and confirm that line is the only place where we do this
• [x] Check the Travis builds and adjust as needed

[ozak]

Any updates on this? While mostly just annoying, it messes up any notebook you create for presentations or courses. Looking at the code it seems to have been added as a hack for fixing the Travis build. A hack in the meantime is to run

import warnings
warnings.filterwarnings("ignore", category=UserWarning)

to eliminate the warning. Not optimal of course.

[smmaurer]

Hi @DavidMStraub and @ozak, thanks for the reminder about this. I just opened a PR to fix it: #135.

Would one of you be able to help with a little testing, to make sure it fixes the issue?

You could either (1) send me a notebook that demonstrates the problem, or (2) install the new Pandana branch and check it on your own machine.

You could install the branch like this:

git clone --branch fix-matplotlib-backend https://github.com/udst/pandana
cd pandana
conda env create -n pandana-test python=3.8 cython numpy clang llvm-openmp jupyter
conda activate pandana-test
pip install -e .

Thanks again for reporting the issue. If things go smoothly the fix should be up on pip/conda next week, along with some other updates.

[ozak]

I was getting the issue via pysal. It seems the issue is resolved in the latest version of pysal-2.2.0.

[knaaptime]

that's because we changed to a lazy import structure, so if you aren't using a handful of specific functions from segregation, then pandana never gets imported. Should be testable following a similar strategy as @jreades earlier in the thread

[smmaurer]

Thanks @knaaptime! I just tried this and was able to reproduce the behavior and also confirm that the new PR fixes it. Sorry for all the problems this caused.

[knaaptime]

no worries, this is great. thanks for the fix :)

[smmaurer]

Resolved in PR #135.