Brainstorm functions inside package

[lwjohnst86]

From what I read of the comments, seems that @AndersAskeland ideas (science-collective/admin#6 (comment)) are a good starting point, with the others agreeing. Specifically points 1 ("Metapackage") and 2 ("guided data analysis setup"), which could actually be merged together a bit.

So, here, let's write out some ideas for individual functions, names for the functions, what type of input they need, and what type of output they give (or actions they do).

Before we even get to actually building them, it really helps to have a conceptual, text description of the functions, otherwise you can easily flounder around with just dealing with random R issues. I speak from experience 😛

[AndersAskeland]

Depending on what exactly we build, there is a bunch of ways we can use the RStudio interface. I found this, that summarises some of the methods we can use.

As for the guided analysis (if that is something we go for), I figure we can use the viewer pane to display HTML - I would then be amazing if one could also connect the HTML progress to the executed R code (this might require Shiny).

[lwjohnst86]

@AndersAskeland Yea, those functions make it so much easier to work within and customize RStudio to do what you want.

For the guided analysis, at least for now we could have code output that people could copy and paste as the final output. Similar to how RStudio's "Data import" works.

[lwjohnst86]

This section of the rOpenSci package dev book (https://devguide.ropensci.org/building.html#interactivegraphical-interfaces) has information on generating code from a workflow steps, like might be done with the guided analysis

[lwjohnst86]

One idea is to convert multiple checklists (PRISMA, CONSORT, STROBE) into machine-readable formats that could be output as Markdown files or created as GitHub Issues.