In our project, we have a directory (let's just call it "/sw/") where we keep everything related to the project. Various scripts, templates, delivered files, git-repos, builds, etc are in subdirectories of /sw/.
This is an old project, and /sw/ have become quite a mess. A lot of it seems totally obsolete, and probably should be removed. It's hard to do this though, since there's always a small risk that it's actually used somewhere.
Things in Git repositories should remain, those are well documented and managed. However, there's a lot of subdirectories and files in /sw/ not under version control, and much of this seems like junk.
Edit to clarify: Any existing git repos under /sw/ would not be a part of the /sw/ repo. For example /sw/develop.git would never be added.
If we should do a cleanup of /sw/, we are considering two strategies for this:
1) Make a folder somewhere called "obsolete", and just move the (presumably) junk files there. We'd perhaps keep a textfile in that directory as a log of what has been moved, and why.
2) In /sw/, initialize a Git repository. Then, each time we do cleanup and something is removed, we would first add and commit it to the /sw/ repository, then remove the files, then commit that change. Then if we discover something should not have been removed, we can just revert that change.
Which one seems like the best/safest strategy? I sometimes hear "keep everything under version control". But is it just making things more complicated in this case? Or would it be a benefit?
Is there some other way than the two points listed above?