Open mniederhuber opened 3 years ago
I think the following process makes sense:
mckaylabunc
organizationSubmodules are relatively simple to operate, but using them does require multiple commands so it might be additional mental overhead?
I am sneaking in to say that submodules are often more trouble than they're worth. In my experience for a published analysis, it doesn't matter whether you can link the pipeline back to a git hash, what you really just want is the source of the pipeline you used embedded into the repo. It makes upgrading the pipeline a little more annoying during an analysis, but it really shouldn't matter that much cause the changes will all get tracked in git & it's trivial to roll them back.
Sincerely,
Someone who suffered using submodules for this purpose early in their PhD. ❤️
Aha ok that's definitely valuable to hear! I haven't used them very much. So then the above would be the same except cloning repo inside analysis dir instead of using submodules.
cloning repo inside analysis dir instead of using submodules.
And doing an rm -rf .git
inside the repo so you avoid recursive versioning.
Does a project repo have to be in the main mckaylabunc
org to use pipelines as submodules?
Personally, I think it will help keep the org clean and easy to navigate if we discourage personal project repos in mckaylabunc
, my fear is that we will end up with a lot of "wounded soldiers" so to speak
That sounds reasonable.
Does a project repo have to be in the main
mckaylabunc
org to use pipelines as submodules?
I don't believe so - it sounds like submodules would be a "if you so dare" situation anyway
Accidentally closed - sorry