Hello!
I am working on a remote computer, where I log into with SSH (and which is available only via company VPN).
I tried running the example from Build an App — Panel v1.5.2 with panel serve --address=<IP> --port=5006 --allow-websocket-origin=<IP>:5006 --dev notebook.ipynb, and it runs without any problems, and work correctly.
However, when I tried to do the same with the matplotlib example from Matplotlib — Panel v1.5.2 (without any changes, downloaded from https://raw.githubusercontent.com/holoviz/panel/main/examples/reference/panes/Matplotlib.ipynb with wget), I get quite strange behavior.
- When I click on ‘Preview with Panel’ button, I get almost empty page - however, when I created a simpler notebook based on the
Matplotlib.iynbone, but withpn.template.EditableTemplate([...]).serverable(), everything renders correctly - When I run Panel from the terminal with
panel serve --address=<IP> --port=5006 --allow-websocket-origin=<IP>:5006 --dev <notebook>.ipynb, I always get almost empty page.
The panel serve ... command does not show any errors or warnings:
$ panel serve --address=<IP> --port=5006 --allow-websocket-origin=<IP>:5006 --dev Matplotlib.ipynb
2024-10-15 13:29:21,976 Starting Bokeh server version 3.6.0 (running on Tornado 6.4.1)
2024-10-15 13:29:21,977 User authentication hooks NOT provided (default user enabled)
2024-10-15 13:29:21,978 Bokeh app running at: http://<IP>:5006/Matplotlib
2024-10-15 13:29:21,979 Starting Bokeh server with process id: 2373293
2024-10-15 13:29:51,279 WebSocket connection opened
2024-10-15 13:29:51,279 ServerConnection created
2024-10-15 13:29:59,477 404 GET /sm/a66f7e8440d7b0538d7c7e64fb20eaeb1b8e7204179c205d57a779774b7b22d9.map (172.18.56.66) 0.70ms
When I inspect the almost empty page from panel serve ... I see the following problems and errors:
-
attempt at accessing the localhost (127.0.0.1), even though I have provided the
--allow-websocket-originparameter (question: is there parameter that I should have provided in addition)?(3)(+0002033): Error: HTTP request to http://127.0.0.1:23119/connector/ping rejected with status 0 -
Muuri being not defined
Uncaught ReferenceError: Muuri is not defined at Matplotlib:311:16 -
undefined value error that is most probably caused by the previous error:
bokeh.min.js?v=de958…62521f0895deb32:164 Error rendering Bokeh items: TypeError: Cannot read properties of undefined (reading 'on') at eval (eval at _render_script (panel.min.js?v=03554…3f4c6edba4:225:6767), <anonymous>:36:6) at M.run_script (panel.min.js?v=03554…3f4c6edba4:225:2895) at M.render (panel.min.js?v=03554…3f4c6edba4:225:3719) at M.render_to (bokeh.min.js?v=de958…1f0895deb32:223:512) at M.build (bokeh.min.js?v=de958…1f0895deb32:223:830) at bokeh.min.js?v=de958…1f0895deb32:220:320 at async l (bokeh.min.js?v=de958…1f0895deb32:220:198) at async bokeh.min.js?v=de958…1f0895deb32:220:408 at async k (bokeh.min.js?v=de958…f0895deb32:164:1098) k @ bokeh.min.js?v=de958…62521f0895deb32:164
From what I found, Muuri is a JavaScript library to create responsive, sortable, filterable and draggable layouts.
Why it is not a problem for hvPlot, but is a problem for Matplotlib? Why it (sometimes) works with ‘Preview with Panel’, but not with panel serve --address?
How can I try to solve this problem? Is there some command line argument that I should have provided, or perhaps some configuration option?
List of relevant packages installed:
- bokeh==3.6.0
- holoviews==1.19.1
- hvplot==0.11.0
- ipykernel==6.29.5
- ipympl==0.9.4
- ipython==8.27.0
- ipython-genutils==0.2.0
- ipywidgets==8.1.5
- ipywidgets_bokeh==1.6.0
- jupyter-events==0.10.0
- jupyter-lsp==2.2.5
- jupyter-server-mathjax==0.2.6
- jupyter_client==8.6.3
- jupyter_core==5.7.2
- jupyter_server==2.14.2
- jupyter_server_terminals==0.5.3
- jupyterlab==4.2.5
- jupyterlab_pygments==0.3.0
- jupyterlab_server==2.27.3
- jupyterlab_widgets==3.0.13
- matplotlib==3.9.2
- matplotlib-inline==0.1.7
- panel==1.5.2
- pyviz_comms==3.0.3
- watchfiles==0.24.0