Difficulty Running UI tests

Panel
1

I’m trying the development environment setup and test running for the first time. I followed all the code example steps in the developer guide other than setting up the Jupyter extensions. Then I went onto the test guide I was able to run the unit tests no problem but when I tried running the UI tests with:

doit test_unit

Almost all the tests failed with this error:

image
image3119×1029 161 KB

For that css file path I can see a tabulator_fast.min.css in that directory but not the tabulator_simple.min.css the test is apparently looking for. Not sure if I missed some build or bundling step I was supposed to have done…

1 Like
2

We want to improve this experience, but haven’t got around to it yet.

Can you do the following:

  • SETUPTOOLS_ENABLE_FEATURES=legacy-editable pip install -e . (Just to be sure)
  • panel bundle --all
2 Likes
3

Thanks @Hoxbro ! That’s working now.

1 Like
4

@Owen . If you want to contribute, then a contribution to the developer guide documenting how to get the UI tests working would be very valuable.

I’m leading a sprint at EuroPython next weekend. And it would really help me if the developer guide was in a better shape :smile:

5

I was trying to setup the development environment in editable mode to run the unit tests and ui tests; and got the following error “‘MathSpec’ is not defined”. It is a conda.model package that seems to not be available to pyctdev. Not certain why. I’m new to conda and installed the full Anaconda on macos to try to get through the dev setup. I looked to see if I could pip3 install conda, and see the versions have been pulled form PyPi. Any pointers? Thanks.

doit develop_install -c pyviz/label/dev -c conda-forge -c bokeh -o build -o tests -o recommended

File “/panel/.venv/lib/python3.11/site-packages/pyctdev/_conda.py”, line 432, in _join_the_club
ms = MatchSpec(new)
^^^^^^^^^
NameError: name ‘MatchSpec’ is not defined

(panel_dev) (.venv) panel %

1 Like
6

Sorry this is in such bad shape. We were hoping to have migrated away from pyctdev to a more standardized workflow based on hatch by now but that hasn’t happened. I’ll try to investigate shortly.

1 Like