From 6682b04a2779b8e28f5a1d15b12c48679c37a00b Mon Sep 17 00:00:00 2001 From: fireboy Date: Tue, 9 Nov 2021 01:48:01 +0000 Subject: [PATCH] --- ..._does_adjust_need_a_filepointer_at_all__63__.mdwn | 12 ++++++++++++ 1 file changed, 12 insertions(+) create mode 100644 doc/forum/Why_does_adjust_need_a_filepointer_at_all__63__.mdwn diff --git a/doc/forum/Why_does_adjust_need_a_filepointer_at_all__63__.mdwn b/doc/forum/Why_does_adjust_need_a_filepointer_at_all__63__.mdwn new file mode 100644 index 0000000000..542ed3ce98 --- /dev/null +++ b/doc/forum/Why_does_adjust_need_a_filepointer_at_all__63__.mdwn @@ -0,0 +1,12 @@ +I'm trying to understand why `git annex adjust` changes any symlinks to pointers. All the "pointer file" contains is the git annex object path with the hash, which the symlink also contains. Is there a reason we can't leave the branch untouched and just change the local configuration to track which files are to be adjusted as locked/unlocked? I imagine with this model, we don't even need an adjust command, and we could simply upgrade the old `lock` and `unlock` commands. + +``` +git annex unlock file1 ## Adds file1 to list of files to run the "adjust" smudge/clean filter for. This data could be stored somewhere in .git/annex, but is completely local and does not need to be synced or protected anywhere. +git annex lock file1 ## Removes file1 from list of files to run the adjust on. +``` + +What is the advantage? + +For one, we don't have extra branches that are potentially confusing. It also simplifies the user facing surface area which is great to reduce user-causing-bugs. + +Thoughts?