Run from within a Jupyter Notebook

When the following code is run from a demo.py file or from the Python interpreter, it runs as expected, collecting a new dataset. However, when run from within a Jupyter Notebook, it complains that KeyError: u'dallinger_email_password is not a valid configuration key'.

import dallinger

experiment = dallinger.experiments.Bartlett1932()
data = experiment.collect(
    app_id=u"3b9c2aeb-0eb7-4432-803e-bc437e17b3bb",
)
print(data.networks.list)

Test files:

demos.zip

[suchow]

@MatthewWilkes I don’t see any issue with using your fork. The pexpect license is permissive, and so if your contribution to pexpect is licensed under the same terms, it shouldn’t pose any problems.

With respect to NGS2, we promised that all the code produced as part of the program would be open source and permissively licensed, not that it’s specifically MIT licensed; this meets those requirements.

[MatthewWilkes]

I’ve added PR #736 that works around this. Modifying pexpect turned out to be a rabbithole, it’s not possible to support its functionality adequately well enough to make something useful. A lot of the same fixes needed were in common with @jessesnyder #707 branch, which makes sense as they’re both about getting useful information from subprocesses when not using a terminal.