Jupytext: Is the Rmd output a valid RMarkdown notebook?

[choldgraf]

This might be an ill-posed question, but I am wondering whether the two-way conversion with RMarkdown formats totally matches the specification of RMarkdown notebooks. E.g., could I take any random RMarkdown notebook and use jupytext to convert it into a Jupyter Notebook, and vice versa? Are there any edge cases or pain points where this wouldn't work? This might be a nice goal for the RMd support, since that would effectively bridge the document formats of Jupyter Notebooks and R documents...

[mwouts]

Hello @choldgraf , well that's how Jupytext started... see issue #1 😄

I would say that the conversion works rather well, at least by the number of issues I am seeing here.

I think on this subject @grst may have more experience than me. He helped me a lot getting this started. He also has experience with the .nb.html format (R notebooks with outputs, the Rmd document is embedded inside). And he has practice of using both Jupyter and knitr to render the notebook, see his reportsrender project. So he'll better know about the edge cases - which I'd be willing to fix anyway!

[grst]

Hi @choldgraf,

I actually use it in production for a while to write python notebooks in jupyter lab and generating HTML reports with rmarkdown and reticulate. (See also, as @mwouts already mentioned, the reportsrender project. Especially the section about executing the same Rmd file with both nbconvert/papermill and rmardown. )
Also I never had issues opening R Rmd files and executing them with the irkernel.

So far, everything works like a charm!
The only caveat is that jupyter cell magics don't work. The Rmd file will still be valid, though. The line containing the cell magic will just be commented out (and, therefore, might break the notebook when executing it with rmarkdown).

[mwouts]

Chris, I think we've answered your question here, so I'll close this issue. Still, feel free to reopen if there is something we did not answer!