I frequently use jupyter nbconvert to generate HTML pages from notebooks. Some of my notebooks use the Plotly library.
Since the release Plotly 6, by default the Plotly plots generate only one output, of type "application data". I believe this is a sensible choice as it makes the notebook much lighter (previously, a plot would generate multiple outputs: the same json data, plus an HTML output that would contain both the plot data and the JS Plotly library).
Now the problem that I encounter is that, while Jupyter Lab is able to display the plot, using I believe the Plotly extension for Jupyter Lab, Jupyter NBconvert is not able to generate an HTML output from the application data. It just ignores the application data and issues this warning:
UserWarning: Your element with mimetype(s) dict_keys(['application/vnd.plotly.v1+json']) is not able to be represented
I know that a workaround is to generate the plot directly in the HTML format (see plotly/plotly.py#5365 for a MRE and the workaround). Yet I was wondering if it was possible / planned to get the Jupyter Lab extension generate the HTML output from the application data?
While this question initially happened in the context of Plotly, I am actually interested in a general approach, as I also maintain a package that creates HTML outputs (ITables) and had been thinking of transitioning to embedding only the data in the cell outputs.