From de4d50e9d09ccd2518ac5e0c5e5f1b1a17a824ce Mon Sep 17 00:00:00 2001 From: Joey Hess Date: Fri, 17 Feb 2017 12:57:55 -0400 Subject: [PATCH] thoughts --- ...er_setup_for_remote_worktree_update_on_push.mdwn | 13 +++++++++++++ 1 file changed, 13 insertions(+) diff --git a/doc/todo/simpler_setup_for_remote_worktree_update_on_push.mdwn b/doc/todo/simpler_setup_for_remote_worktree_update_on_push.mdwn index 8d25f44b94..4f1ffba652 100644 --- a/doc/todo/simpler_setup_for_remote_worktree_update_on_push.mdwn +++ b/doc/todo/simpler_setup_for_remote_worktree_update_on_push.mdwn @@ -40,6 +40,19 @@ adjusted branches? Could install a post-update hook, that runs a git-annex command that checks for updateInstead, and emulates its behavior, handling direct mode and adjusted branches. +To support direct mode repos on removable drives w/o execute bits, +could make sync check local remotes and run the equivilant action as the +hook would run. + +To fully emulate updateInstead, the post-update hook +should abort if the tree is unclean or if there are merge conflicts. +But, in a direct mode repo, the only way the user will likely resolve such +a situation is git-annex sync/merge, so the hook could just run git-annex +merge instead of trying to fully emulate regular updateInstead behavior. +Similarly, in an adjusted branch, the push will update master, and git +annex sync/merge is what the user will likely do. Although they could +choose to reset changes to the tree. + --- Potential least surprise violation: