about summary refs log tree commit diff
path: root/Documentation/RelNotes/1.6.1.2.txt
diff options
context:
space:
mode:
authorVincent Ambo <Vincent Ambo>2020-01-11T23·36+0000
committerVincent Ambo <Vincent Ambo>2020-01-11T23·36+0000
commit1b593e1ea4d2af0f6444d9a7788d5d99abd6fde5 (patch)
treee3accb9beed5c4c1b5a05c99db71ab2841f0ed04 /Documentation/RelNotes/1.6.1.2.txt
Squashed 'third_party/git/' content from commit cb71568594
git-subtree-dir: third_party/git
git-subtree-split: cb715685942260375e1eb8153b0768a376e4ece7
Diffstat (limited to 'Documentation/RelNotes/1.6.1.2.txt')
-rw-r--r--Documentation/RelNotes/1.6.1.2.txt39
1 files changed, 39 insertions, 0 deletions
diff --git a/Documentation/RelNotes/1.6.1.2.txt b/Documentation/RelNotes/1.6.1.2.txt
new file mode 100644
index 0000000000..be37cbb858
--- /dev/null
+++ b/Documentation/RelNotes/1.6.1.2.txt
@@ -0,0 +1,39 @@
+GIT v1.6.1.2 Release Notes
+==========================
+
+Fixes since v1.6.1.1
+--------------------
+
+* The logic for rename detection in internal diff used by commands like
+  "git diff" and "git blame" has been optimized to avoid loading the same
+  blob repeatedly.
+
+* We did not allow writing out a blob that is larger than 2GB for no good
+  reason.
+
+* "git format-patch -o $dir", when $dir is a relative directory, used it
+  as relative to the root of the work tree, not relative to the current
+  directory.
+
+* v1.6.1 introduced an optimization for "git push" into a repository (A)
+  that borrows its objects from another repository (B) to avoid sending
+  objects that are available in repository B, when they are not yet used
+  by repository A.  However the code on the "git push" sender side was
+  buggy and did not work when repository B had new objects that are not
+  known by the sender.  This caused pushing into a "forked" repository
+  served by v1.6.1 software using "git push" from v1.6.1 sometimes did not
+  work.  The bug was purely on the "git push" sender side, and has been
+  corrected.
+
+* "git status -v" did not paint its diff output in colour even when
+  color.ui configuration was set.
+
+* "git ls-tree" learned --full-tree option to help Porcelain scripts that
+  want to always see the full path regardless of the current working
+  directory.
+
+* "git grep" incorrectly searched in work tree paths even when they are
+  marked as assume-unchanged.  It now searches in the index entries.
+
+* "git gc" with no grace period needlessly ejected packed but unreachable
+  objects in their loose form, only to delete them right away.