Need help in trying to configure Git repositories for users who are using RStudio

[Rod-at-DOH]

🏷️ Discussion Type

Question

Body

I am one of the GitHub Administrators where I work. I'm a developer, focusing primarily on Microsoft .NET/C#. But where I work other programming languages are used. Two of them are SAS and R. I am familiar with SAS and R by reputation only. Various groups here have used SAS for many years and are starting to use R, in RStudio, more. There's a major project to migrate all the SAS code into R.

For me, the hardest thing is trying to figure out how the SAS/R people have used SAS and R for all the years they've been writing scripts in both languages. What's complicated this is the project managers who are involved, will derail any attempt I've made to try to understand how the users have approached configuring the R projects. Once I ask a clarifying question, the PMs drive the discussion elsewhere. This is causing other problems, but the bottom line is I'm no closer to understanding how and why the users configure their R projects the way they do, then I was weeks ago.

So, I've come here, in hopes of gaining some insights. Here's what little I know. I know that they've been developing SAS and R projects on network (Windows) shares. I get the impression (which I've never been able to verify) that they'll build a large and deep hierarchy of SAS/R projects. I've no idea how large nor how deep. Also, I have no idea, no matter how many times I've asked, if any of these individual SAS/R projects are related, or if they're only in this folder structure because it was convenient to do it that way, rather than putting the projects in separate folders rather than embedded folders. I get the impression that one group of people will work on one project, but they are concerned that other groups do not see what they're doing, even if they are in embedded folders. That concern makes no sense to me, because anyone who has access to any folder, should have access to sub-folders, etc. But here I'm not sure I'm understanding it correctly. Maybe they do have separate, base folders with sub-folders for other SAS/R projects which are all related. But like I said, this is one of those questions that even when I bring up a clarifying question, the discussion is forced in another direction, so I never get this question answered.

I hope that this is enough to help you understand the predicament I am in. And I'm hoping there are some people here who have SAS or R repos who also have experience with people writing SAS and R code who know how to configure these repos in a way that my users will be somewhat happy with and conform, at least somewhat, to how they've worked for many years.

[P-r-e-m-i-u-m]

Yo @Rod-at-DOH — I feel for you man. PMs derailing technical discussions is a universal pain 😂
But real talk, you're asking the wrong question. You don't need to understand how SAS/R people configured their projects for 20 years. You need to know what they actually need from Git to not revolt.
Here's the reality:
SAS/R people don't think in repos. They think in scripts, datasets, and "where did I save that .csv." The folder hierarchy you're seeing? It's probably organic chaos from 15 years of "just save it here for now."
What they actually care about:
"Can I still open my project the same way?" (RStudio projects map to folders)
"Will my paths break?" (R uses relative paths, moving folders = pain)
"Can other teams see my code?" (they want isolation, not secrecy — big difference)
The config that actually works:

repo-root/
├── team-a-project/
│ ├── data/ (gitignore this)
│ ├── scripts/
│ ├── outputs/ (gitignore this)
│ └── project.Rproj (RStudio project file)
├── team-b-project/
│ └── ...
└── shared-libs/ (if they actually share code)

One repo per team/project. Not one mega-repo with embedded folders. That's what they're afraid of — Team A doesn't want Team B's commits showing up in their history.
Gitignore everything they don't need versioned:

R stuff

.Rproj.user/
.Rhistory
.RData

SAS stuff

*.sas7bdat
*.sas7bcat

Data

data/
outputs/
*.csv
*.xlsx

The PM problem: Stop asking "how do you configure repos?" Ask "what breaks your workflow?" Then build Git around that.
If you want to talk this through live — I run a weekly builds community where devs share what they're building and stuck on. No corporate PMs allowed 😂
Week W17 is open: github.com/P-r-e-m-i-u-m/weekly-builds-community
Drop your Git config there, I'll help you figure out the repo structure that won't make your SAS/R people quit.

[Rod-at-DOH]

Hey, this is the kind of feedback that I need! I am concerned about the data being in the "repo". I get that data/ and outputs/ can be ignored. But I'm concerned about the size of some of the data. I've had some people tell me that their .csv files are >100MB in size, which as you know won't go into a Git repo, unless we pay for LFS, which isn't going to happen. Of course, as you say, we could ignore those by putting them into the Gitignore file. However, I think a few of these people have PCs with very limited resources. So, they might not even be able to tolerate those files on their systems, which is probably why they've always done development on network shares. I've got to look into that.

I'll check out your Git community.