Test functions instead of cells

[pylang]

Hi. Is it possible to read tests from regular functions and class names? This would alleviate coding each test to a separate cell and allow the use of traditional unittests.

[zonca]

actually the purpose of this plugin is exactly to avoid the need to create functions, without functions is a lot easier to debug and inspect what is going on.

…

On Fri, Nov 24, 2017 at 12:34 AM pylang ***@***.***> wrote: Hi. Is it possible to read tests from regular functions and class names? This would alleviate coding each test to a separate cell and allow the use of traditional unittests. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#26>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAXYctraMf7Puzic2PWSP3aqnT6SaPCLks5s5gEhgaJpZM4QpPmI> .

[pylang]

And I suspect easier to code. For example, you can parse the notebook cell source keys from json, access the cell name and remaining cell code.

I don't mind this idea of running tests in cells. However, using functions helps maintain local scopes. It also does not compel the user to defect from a standard, functional way of writing tests. Consider writing a simple test function:

test_foo():
     assert foo() == expected

Someone familiar with unit tests in python understands what that is - a test on some object called foo. This test can be run in jupyter, or copied/pasted into a module and run by many test runners, i.e. unittest (with some modification), nose and pytest. Here, the tests are restricted to notebook cells.

Supporting tests as cells is fine and interesting. I would only recommend extending this plugin to support an extant convention of writing traditional tests. This improvement might be promising in attacking a bigger issue - that is not being able to run pytest inside a notebook

[zonca]

ok, supporting both would be fine, if it doesn't complicate the code too much. Would you like to contribute this improvement as a pull request?

…

On Fri, Nov 24, 2017 at 9:55 PM pylang ***@***.***> wrote: And I suspect easier to code. For example, you can parse the notebook cell source keys from json, access the cell name and remaining cell code. I don't mind this idea of running tests in cells. However, using functions helps maintain local scopes. It also does not compel the user to defect from a standard, functional way of writing tests. Consider writing a simple test function: test_foo(): assert foo() == expected Someone familiar with unit tests in python understands what that is - a test on some object called foo. This test can be run in jupyter, or copied/pasted into a module and run by many test runners, i.e. unittest (with some modification), nose and pytest. Here, the tests are restricted to notebook cells. Supporting tests as cells is fine. I would only recommend extending this plugin to support an extant convention. This improvement might attack a bigger issue - that is not being able to run pytest inside a notebook. pytest-dev/pytest#2268 <pytest-dev/pytest#2268> — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#26 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAXYcsPr6zwXvF2kd05bbhhyGXE7FR1aks5s5y1YgaJpZM4QpPmI> .

[pylang]

Thanks. I cannot contribute a pull request at the moment, but perhaps I may in the future. I admit the learning curve is steep to make a contribution at the moment. In the meantime, those interested on continuing this idea are welcome.