mirrors/git-annex
2
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Browse source
This commit is contained in:
jg123h12jh3y12g3y 2023-03-22 09:47:31 +00:00 committed by admin
parent a40a36495f
commit fc581d0c20
1 changed files with 11 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

11
doc/forum/Clarifications_about_how_to_work_with_git-annex.mdwn Normal file
View file

@ -0,0 +1,11 @@
Hello. I am a kinda-new git-annex user. I think git-annex is very difficult to understand. Thus, in this topic, I would like to write out my explanation of how git-annex works and how I can work with it, so that you can tell me if I am misunderstanding something.
I am a machine learning engineer. I use git-annex to store code together data (large images) and artefacts (large images and model weights). Basically, I am trying to use git-annex as normal git, but with the ability to store large binary files. There is a central bare repository and a bunch of non-bare repositories cloned from it on different computers. Usually, the non-bare repositories connect only to the central bare repository, but sometimes I push/fetch between non-bare repositories directly.
I can fetch/push/merge/pull the normal branches I work in just as I do with normal git (without annex). However, the git-annexed files aren't fetched/pushed this way because the branch actually contains only symlinks, not the files themselves. I have disabled all the automatic mergin, pulling, etc. functionality of git because I like to have total control. So, I merge or rebase everything manually when there is a need for it. And I never use pull.
In each repository, there is a branch called git-annex. It contains some metadata that git-annex uses. While in a repository repo1, I can do `git annex sync --only-annex --no-content --no-commit --no-pull --no-push --no-resolvemerge repo2` and git-annex will use some magic (consisting of pushing, pulling, and some kind of automatic merging) to sync the git-annex branch of repo1 and repo2, i.e. it will make them contain the same metadata. The options `--no-commit --no-pull --no-push --no-resolvemerge` are needed to disable the dark magic that is useful for casual users but not for software developers who use git-annex as a git addon. The option `--only-annex` prevents git-annex from creating "synced" branches which are, as far as I understand, another piece of dark magic useful for casual users but not for software developers. If I want, I can remove the `--no-content` flag and git-annex will also download and/or upload the annexed data (does it affect only the data available in the current branch? or is it all data? i'm not sure). This is the only command I need to know to sync the git-annex branch. Supposedly, it's possible to do via normal git fetches, pushes, merges, and maybe pulls, but I don't know how to do that.
The actual annexed data is stored somewhere in the .git directory. I don't need to worry where. What I need to know is that I can use `git annex copy`, `git annex get`, and `git annex sync --only-annex --no-commit --no-pull --no-push --no-resolvemerge otherrepo` with appropriate paths to copy the annexed data between repositories.
Ok, so I can use `git annex sync` with a bunch of flags to sync the git-annex branch, I can use `git annex sync`, `git annex get`, `git annex copy` to copy the data around, and I don't need the synced branches. Is my understanding correct?