Data Flow for RStudio projects


#1

I often thought about a data flow viewer fo organize several scripts or code chunks within a project.
The idea is not new to R http://r.analyticflow.com/en/ and of course there are many many good examples of non r solutions out there (also for statistics/machine learning).

However, with a bit of thought, this might be a nice helper and could offer lots of possibilities

  • visualise workflow also for collaboration and presentations (also like Anyone else interested in a minimap for Rstudio?)
  • encouraging modular development, clean APIs between tasks and maybe also strong typing
  • automated caching between tasks
  • indicate parallel tasks
  • switching between languages like R + Python in IDE
  • several things like comments, codeprevies, other metainformations could come up as tool tip
  • etc

Were / are there thoughts about integrating such a feature into the IDE?


#2

I also find this idea intriguing. In the age of htmlwidgets and add-ins, I don’t even think that it necessarily is dependent on changes to the IDE.

There exist packages for visualizing package dependencies and function dependencies. For example: http://datastorm-open.github.io/DependenciesGraphs/

I have been studying the draft chapters in the Advanced-R book, and have been getting unreasonably excited about the possibilities of the meta-programming tools in R and rlang/tidyeval. A fun project I would like to experiment with would be a version of source() that constructs an network visualization instead of evaluating code, and pops it up in an htmlwidget. (level 2: if you make changes in the htmlwidget, would it be possible to feed that back to code changes in Rstudio?)


Data flow GUI for Shiny development?