idea

doc/todo/hide_missing_files/comment_2_bdae4301aa8ddc6621d4eb2612627069._comment

@@ -0,0 +1,37 @@
+[[!comment format=mdwn
+ username="joey"
+ subject="""comment 2"""
+ date="2018-10-14T05:08:18Z"
+ content="""
+The blocker to implementing this in views (or adjusted branch) has been
+how to efficiently update the tree after a get/drop. Doing it after every
+file is certianly too slow and also would create lots of unused tree
+objects in git that would linger until gced.
+
+And less of a blocker, but still kind of a problem is, how to run `git
+annex get` on a file if its symlink is hidden due to it not being available?
+
+Occurred to me tonight that this could be done in `git annex sync --content`,
+at the end. That way the tree update would only be done once, after however
+many gets/drops. And sync could look at the original branch when finding
+content to transfer, so would avoid the other problem too.
+
+One of my use cases for wanting this is to use it on my phone, to load up
+podcasts I want to listen to on a trip. One way I could do this is
+to use metadata to tag those podcasts with a tag that is preferred content
+for the phone. Then `sync --content` run on the phone
+would download them, and update the visible tree.
+
+The assistant could also update the visible tree at the end of a sync run.
+
+Could `sync --content` cause an update of a remotes visible tree too?
+It would need an addtion to the P2P protocol, but seems worth
+doing for symmetry.
+
+(For users who don't want to use sync but want to git-annex get/drop
+manually, they could check out the real branch temporarily, run whatever
+there, and then run a command to get back into the view where
+missing files are hidden. Of course this could be slow when the repo has a
+lot of files. Especially in v6 mode where checkout runs git-annex once per
+file.)
+"""]]