Closed philippjfr closed 3 years ago
Have to figure out the tests but locally this now works in classic notebook and jlab when following the dev instructions. Ping @mattpap and @bryevdv for testing.
Actually, the nbextension doesn't seem to work yet.
Fixed, should all work now.
Tests finally passing as well.
I have publish 3.0.0-dev1
to NPM and PyPI. Seems to moslty be working:
A few notes:
push_notebook
does not work (no comms available)Well, actually looking at the console, I think maybe it is not using the extension for me, but instead be falling back to the bare JS exec that classic uses?
cc @philippjfr
@jasongrout Do you have any ideas about the above? It seems I am seeing that error from the installed pre-built extensions, and others are not. But I think I have everything installed that is mentioned in other (old) threads I can find about that message:
(230_39) ❯ jupyter labextension list
JupyterLab v3.0.12
/Users/bryan/anaconda/envs/230_39/share/jupyter/labextensions
@bokeh/jupyter_bokeh v3.0.0-dev1 enabled OK (python, jupyter_bokeh)
@jupyter-widgets/jupyterlab-manager v4.0.0-alpha.1 enabled OK (python, jupyterlab_widgets)
~/work/bokeh/examples/howto bryanv/hasprops*
(230_39) ❯ conda list ipywidgets
# packages in environment at /Users/bryan/anaconda/envs/230_39:
#
# Name Version Build Channel
ipywidgets 8.0.0a4 pypi_0 pypi
~/work/bokeh/examples/howto bryanv/hasprops*
(230_39) ❯ pip install jupyterlab_widgets
Requirement already satisfied: jupyterlab_widgets in /Users/bryan/anaconda/envs/230_39/lib/python3.9/site-packages (2.0.0a1)
How can I debug this?
OK seems like it was the 2.0.0a1 version of jupyterlab-widgets that the pip install --pre jupyter_bokeh
brought it that was busting everything for me
OK seems like it was the 2.0.0a1 version of jupyterlab-widgets that the
pip install --pre jupyter_bokeh
brought it that was busting everything for me
Yes, that --pre
switch bitten me too before. Glad you figured it out.
I took jupyter_bokeh
for a quick test run and it seems to work. Thanks for updating it!
Thanks @jasongrout. One quick question, for a conda-forge recipe should we include jupyter-packaging as a host
dependency or will it skip invoking the build step as long as the package on PyPI has already bundled it?
Thanks @jasongrout. One quick question, for a conda-forge recipe should we include jupyter-packaging as a
host
dependency or will it skip invoking the build step as long as the package on PyPI has already bundled it?
I think because setup.py imports it, it needs to be a build dependency, like here:
Another attempt at this, this time by reinitializing the project and then readding the old code.