json -> yaml first, and use pandoc scholar ?

[jcolomb]

I see that the different outputs are created from the spreadsheet information.

I would take some time to think if on the long term, it would not be better to produce the yaml version (i,e, a list in R) and then use pandoc/pandoc scholar to create the other outputs. This would probably solve #31 and many other thing that might come once we try to incorporate more infos (orcid number, funding, non-author contributors,...)

[marton-balazs-kovacs]

Hi!

Do you mean that the app would take a list or a YAML file as the input instead of the spreadsheet? Or rather the input spreadsheet would stay, but upon upload, it would be translated to list (YAML), and then the text output would be rendered from that?

We were thinking about the second option that I mention, but I do not see a big advantage as long as the input remains in a spreadsheet format (please correct me if I am missing something). We like this format because it is easily understandable for non-tech-savvy users, and it can be shared among coauthors so everyone can add their own information.

[jcolomb]

Both are possible, but if we keep with option 2 (spreadsheet is the input), the idea would be to simplify the workflow once more information are added (orcid number for instance), or in case some outputs would need to change:

instead of writing 3 codes for the different outputs, one would only need to produce the yaml and then create other outputs from the yaml via pandoc (latex, epub, docx, html, jats). This could be done via Rmarkdown (see https://github.com/open-science-promoters/pandoc-scholar/blob/master/example/rmarkdown-scholarpandoc.Rmd).

It also means that if the jats recommandation would change, we would only need to update pandoc-scholar and not update all the code.