Closed mbcann01 closed 6 years ago
After you've added this functionality, change the version number for bfuncs to 0.4.
Add release number: https://github.com/brad-cannell/bfuncs/releases
Maybe it works using (partly) a network diagram where the root node is the project root, the other nodes are files (color coded by type, I.e., data manipulation or analysis or presentation) and the edges are data (files, but maybe other data too).
Hovering over the nodes and edges shows ploty-style metadata.
Sent with GitHawk
Shiny app would be viewable offline, but still hide the code from the user and allow for arbitrary notes.
Sent with GitHawk
But, it isn’t viewable on a GitHub repo. However, as long as it’s just metadata, I can probably link the the shiny app hosted on shiny apps.io from Github
Sent with GitHawk
What do I do about automatically tracking changes to data if I run a makefile?
For example, data frame x has 63 obs and 20 variables. I currently save that as a comment. Then, if I make a change I can see if data frame x still has 63 obs or not and if that's desirable or not.
In Reproducible research with r and r studio, Gandrud talks about using simple text files as much as possible. The reason is to future proof the work. Text files are unlikely to change much and are likely to be able to be used for many years to come. This is a great point. However, going this route also limits your use of some really nice tools like R notebooks and Shiny. What's the right balance?
Closed this issue because I created an entire package for this functionality -- pathfinder.
Overview:
Need to come up with a better method for notes/documentation/file dependencies for analysis projects.
What this isn't for:
This isn't really intended to be useful for R package development (although, maybe it ultimately will be). It is intended for research projects
General notes that are only intended for me to read (e.g., "left off at line 157" or "Think about using a tree model instead")
What I've already tried:
Using a text file is very convenient, but not very dynamic. It likes features like tables and charts. And, it is not ideal for sharing/collaborating.
Google docs has a lot of features and makes sharing/collaborating easy, but isn’t very convenient. I have to find it, open it (can't open directly in RStudio), and go back and forth between RStudio and Web browser. Additionally, they aren't dynamic in the sense that changes I make to R files and data aren't automatically pushed to the Google doc. They just don't feel like a very native solution.
Github projects has some really nice features; however, it isn’t available offline. Is this a big deal? How often am I really doing work without an internet connection? Another issue is that many of the people I work with are not accustomed to, or comfortable with, using Github.
Specifics:
Packages that may be involved: