summaryrefslogtreecommitdiff
path: root/doc
diff options
context:
space:
mode:
authorGravatar Joey Hess <joey@kitenet.net>2011-02-27 12:17:19 -0400
committerGravatar Joey Hess <joey@kitenet.net>2011-02-27 12:17:19 -0400
commitb14ad6338150243b7f60d0c59aeebf7b0c6e1d03 (patch)
tree449e2f70ce7fa07fb7a7e13afb3d770429956dec /doc
parent22c8dfcd3fb8e052b2c3ae10da6ebc862d6e6efc (diff)
responses
Diffstat (limited to 'doc')
-rw-r--r--doc/bugs/add_range_argument_to___34__git_annex_dropunused__34___.mdwn9
-rw-r--r--doc/bugs/git_annex_migrate_leaves_old_backend_versions_around.mdwn18
2 files changed, 26 insertions, 1 deletions
diff --git a/doc/bugs/add_range_argument_to___34__git_annex_dropunused__34___.mdwn b/doc/bugs/add_range_argument_to___34__git_annex_dropunused__34___.mdwn
index 97b04e292..dc9b7acda 100644
--- a/doc/bugs/add_range_argument_to___34__git_annex_dropunused__34___.mdwn
+++ b/doc/bugs/add_range_argument_to___34__git_annex_dropunused__34___.mdwn
@@ -7,3 +7,12 @@ A range argument, such as `1-1845`, possibly combined with other argument types
I work around this lack as I want to drop all unused files anyway by something like this:
git annex unused | grep -o -P "^ [0-9]+" | xargs git annex dropunused
+
+> It's designed to be used with `seq`. There's an example in the
+> [[walkthrough]], and of course multiple seq calls can be used to
+> specifiy multiple ranges. So:
+
+ git annex dropunused `seq 1 9` `seq 11 1845`
+
+> I don't see adding my own range operations to be an improvement worth
+> making; it'd arguably only be a complication. --[[Joey]] [[done]]
diff --git a/doc/bugs/git_annex_migrate_leaves_old_backend_versions_around.mdwn b/doc/bugs/git_annex_migrate_leaves_old_backend_versions_around.mdwn
index 2fed00be7..c32b763a2 100644
--- a/doc/bugs/git_annex_migrate_leaves_old_backend_versions_around.mdwn
+++ b/doc/bugs/git_annex_migrate_leaves_old_backend_versions_around.mdwn
@@ -1 +1,17 @@
-`git annex migrate` leaves old, unlinked backend versions lying around. It would be great if these were purged automatically somehow.
+`git annex migrate` leaves old, unlinked backend versions lying around. It
+would be great if these were purged automatically somehow.
+
+> Yes, this is an issue mentioned in the [[walkthrough#index14h2]].
+>
+> Since multiple files can point to the same content, it could be that
+> only one file has been migrated, and the content is still used. So
+> the content either has to be retained, or an operation as expensive
+> as `git annex unused` used to find it something else still uses it.
+>
+> Rather than adding such an
+> expensive operation to each call to migrate, I focused on hard-linking
+> the values for the old and new keys, so that the old keys don't actually
+> use any additional resources (beyond an extra inode).
+>
+> This way a lot of migrations can be done, and only when you're done you
+> can do the more expensive cleanup pass if you want to. --[[Joey]]