Open axkibe opened 4 months ago
Hey, thanks for the suggestion! I'll mark this as a possible enhancement and put it on our backlog.
In the meantime, if anyone has time to write up a patch to introduce this functionality (and some tests to go with it), we are always appreciative of contributions to the project!
Since I got a database of all LFS stored objects I can just look that through to build a table to get the sizes to get the full keys. It's just an extra step for me. Reading the docs back then I got the idea it's up to the implementation if oids alone or oids+size are used to identify objects and with size it sounded like a cheap way for extra safety against collisions.
I tried before to hack something to git-lfs, but I must say, since it is Go I could not make heads or tails out of it.
Another wish for ls-files I would like to exclude not a path but a specific commit hash (that commit which uploaded a lot of unnessary files) while I can easily get of course all the oids+sizes from the commit, I cannot easily tell, if they arent used elsewhere (unlikely, but possible)
Describe the issue I want to create a list of all LFS files actually in use of all repositories (excluding deleted repositories and aborted commits)
[Someone uploaded a huge amount of silly LFSs file I need to cleanup]
In my LFS system I used the file size as part of the identification of the file.
With
git lfs ls-files- --all -ls
I'm missing that info. Also at the end of the filename it's hard to parse.Ideally would be a --format option like many other git commands haven then I could do: --format="%O-%S".