How do you organize your git repos?

[langkabh]

As mentioned in today's meeting, I'm curious how everyone is organizing their content in and across git repos (especially in a clinical analysis context): Do you use only one giant repo or do you use multiple, purpose-limited repos? Or rather: When do you create a new repo? For each reporting event? For each study? For each clinical project? And how do you make it possible to reuse code across repos? And one more, related question: How do you organize your content within a repo? What folder structure do you have within your repos?

[aassuied-ps]

I do not yet use git for clinical studies, only for transversal activities (package development, documentation, etc). Still I think having one repository per milestone (reporting event) could be heavy. I like the idea to work on a project (compound) basis, but I'm wondering how it could be handled "easilly".

From my own perspective, with no experience expect transversal and considering what we talked about during the previous meetings, here are my thoughts about a study based repository:

• One main branch with the current fully validated content
• One devel for development
• Separated branches to merge on devel (as shown by Kieran last week) if several persons are working on a study at the same time or one branch per milestone.

I think about two options:

1- copy the devel branch to a specific branch for each milestone and keep it

2- merge the devel to the main and create a release tag by Milestone to identify its and always keep a track.

option 1- could be better when several people are working on the study at the same time on several milestones at the same time to avoid people to work on the same files at the same time if we avoid the devel step and go directly to "milestone_n"

To me the ultimate goal would be to have a project (compound) based repository because:

• A lot of common analysis are performed during within the same project between studies
• It could facilitate pool analysis within the same compound