Extend cell metadata standards

[mwouts]

The official documentation says that Metadata fields officially defined for Jupyter notebooks are (...) collapsed, autoscroll, deletable, format, name and tags.

In practice, it would be useful to have standard fields for, at least

• hiding cell input (this conversation mentions hide_input, is that already a standard?)
• hiding cell text output (and possibly, distinguish between printed messages, warnings and errors)
• displaying raw output (i.e. what we can achieve currently with the Markdown display, cf here)

Note I am aware that the current way to apply the same filtering operation on all cells is with templates, but here I am asking for a cell specific filtering. Also, the same conversation explains how to use cell tags to then filter specific cells, but again I'm afraid tags there are not standard (yet?) neither.

[mwouts]

There is a documentation on a jupyter extension that uses hide_input and hide_output metadata at http://jupyter-contrib-nbextensions.readthedocs.io/en/latest/nbextensions/runtools/readme.html. These cell metadata only become effective, however, when used with the conversion template hide_input_output.tpl.

I note that this is a very common demand. See also

• Hide input cells #534
• Edit meta-data to suppress code input and cell numbers for reveal.js jupyterlab/jupyterlab#4100

[williamstein]

@jasongrout shared a bunch of relevant wisdom in comments here

I personally wish the editable metadata flag, which I guess some extensions (and nbgrader?) use, were also documented in the current official docs. Both Jupyter classic and JupyterLab do indeed prevent "editing" of the cell. But the meaning of "editing" a cell is ambiguous, which is why it would be nice to document a choice... since I am implementing this for CoCalc.com, and have to make a choice. For example, I did a quick check, and it seems to me that if a cell has {editable:false} then in Jupyter classic it cannot be deleted, but in JupyterLab it can be deleted. Which should I choose? (I'm choosing the Jupyter classic behavior, which seems more likely to be what end users expect, though it is less flexible.)

[haraldschilly]

IMHO it isn't a good idea to prevent deleting of cells that shouldn't be edited ... i.e. the jupyter lab behavior is more straight forward and has less surprises. That way, overall there are more options available.