Merge branch 'master' of ssh://git-annex.branchable.com

2 files changed, 43 insertions, 0 deletions

diff --git a/doc/design/external_special_remote_protocol/comment_9_15c4cfe064be37cc104dcb6aa049a449._comment b/doc/design/external_special_remote_protocol/comment_9_15c4cfe064be37cc104dcb6aa049a449._comment
new file mode 100644
index 000000000..cc3f7dd23
--- /dev/null
+++ b/doc/design/external_special_remote_protocol/comment_9_15c4cfe064be37cc104dcb6aa049a449._comment
@@ -0,0 +1,14 @@
+[[!comment format=mdwn
+ username="http://joeyh.name/"
+ ip="209.250.56.227"
+ subject="comment 9"
+ date="2014-01-02T00:15:28Z"
+ content="""
+Tobias made some good points:
+
+* git-annex may not be in PATH depending on installation method
+* It would in theory be bad if a special remote ran some git-annex command that used the special remote and ran some git-annex command [...].
+* git-annex would need to tell the special remote what git repo it was being used with.
+
+So, added GETWANTED and SETWANTED. However, if I find myself recapitulating a lot of git-annex's command-line plumbing stuff in this protocol, I will need to revisit this decision and find a better way. Particularly, I narrowly escaped an intractable dependency loop in [[!commit 8e3032df2d5c6ddf07e43de4b3bb89cb578ae048]].
+"""]]
diff --git a/doc/todo/untracked_remotes/comment_3_0d07c5bc8d42f35351c11411eaca88df._comment b/doc/todo/untracked_remotes/comment_3_0d07c5bc8d42f35351c11411eaca88df._comment
new file mode 100644
index 000000000..946eeaa6c
--- /dev/null
+++ b/doc/todo/untracked_remotes/comment_3_0d07c5bc8d42f35351c11411eaca88df._comment
@@ -0,0 +1,29 @@
+[[!comment format=mdwn
+ username="https://www.google.com/accounts/o8/id?id=AItOawl9sYlePmv1xK-VvjBdN-5doOa_Xw-jH4U"
+ nickname="Richard"
+ subject="comment 3"
+ date="2014-01-02T00:26:14Z"
+ content="""
+Regarding 1.: If two untracked repositories are talking to each other, they should not be tracked at all, so I don't see any issue there.
+If an untracked repository communicates with a tracked one, the untracked one should still send updates for the tracked one when synching.
+The solution might really simply be a specific untracked location log distinct from the rest.
+This would even allow merging changes back into the main log if the user decides to track a repository after all.
+
+Regarding pushing to tracking branches: This behavior will change soon and you can override it; see the manpage for `git-config(1)` at push.default.
+
+Location leaks could be solved by passing `00000000-0000-0000-0000-000000000002` as UUID.
+Using that UUID might also be the solution for all untracked repos as it's trivial to special case for this, but:
+* What happens when you switch a known repo to untracked? What happens to its UUID in various logs? Maybe introduce a specific discard log which tries to get rid of all data concerning those UUIDs?
+* What happens when you switch a repo from untracked to tracked? Simply generate (reactivate?) a UUID and switch all local occurences of `00000000-0000-0000-0000-000000000002` to the new UUID?
+
+`git annex drop --from publicrepo` is not allowed to take local copies into account to satisfy `numcopies`, simple as that.
+IMO, this is the only valid approach, as that mirrors the global view from all other repos.
+For all intents and purposes, an untracked repo does not exist.
+
+
+The complement, a read-only repo, would also be very useful.
+Such a repo would hold data, but it would never accept location data of anywhere besides itself and the web remote.
+
+
+Richard
+"""]]