computationalmodelling/nbval

Possible framework alternative

glass-ships opened this issue · 2 comments

Recently, I wrote a python script which uses nbformat and nbconvert to operate as a simple test that a jupyter notebook runs without error. Output comparison is something I'm interested in but haven't added quite yet. It's available in a gitlab repo here.

That said, I'm curious as to why py.test is used here preferentially (ie. is there functionality specific to the pytest suite that outweighs the benefit of something like my script?

Secondary to that, I've noticed in my research that py.test has been deprecated in favor of pytest - is there any motivation (or need) to update the calls made?

It was just how the project started. pytest does bring some nice architecture for selecting and excluding files, though it probably wouldn't be hard to implement something similar. It's also increasingly what all our projects use to run their regular test suite, so we've generally got it installed anyway.

py.test vs pytest shouldn't make much difference. Maybe some tests and docs need to be updated.

Thanks so much for the response! I'll mark this closed and consider both viable options.