Am 05.12.18 um 07:20 schrieb Elijah Newren:
On Tue, Dec 4, 2018 at 7:54 PM Junio C Hamano <gits...@pobox.com> wrote:
Elijah Newren <new...@gmail.com> writes:
Gah, when I was rebasing on your patch I adopted your sentence rewrite
but forgot to remove the "sometimes". Thanks for catching; correction:
-- 8< --
Subject: [PATCH v2] git-rebase.txt: update note about directory rename
detection and am
In commit 6aba117d5cf7 ("am: avoid directory rename detection when
calling recursive merge machinery", 2018-08-29), the git-rebase manpage
probably should have also been updated to note the stronger
incompatibility between git-am and directory rename detection. Update
it now.
Signed-off-by: Elijah Newren <new...@gmail.com>
---
Documentation/git-rebase.txt | 8 ++++++--
1 file changed, 6 insertions(+), 2 deletions(-)
diff --git a/Documentation/git-rebase.txt b/Documentation/git-rebase.txt
index 41631df6e4..ef76cccf3f 100644
--- a/Documentation/git-rebase.txt
+++ b/Documentation/git-rebase.txt
@@ -569,8 +569,12 @@ it to keep commits that started empty.
Directory rename detection
~~~~~~~~~~~~~~~~~~~~~~~~~~
-The merge and interactive backends work fine with
-directory rename detection. The am backend sometimes does not.
+The merge and interactive backends work fine with directory rename
I am not sure "work fine" a fair and correct label, as rename is
always heuristic.
The "directory rename detection" heuristic in "merge" and the
"interactive" backends can take what happened to paths in the
same directory into account when deciding if a disappeared path
was "renamed" and to which other path. The heuristic produces
incorrect result when the information given is only about
changed paths, which is why it is disabled when using the "am"
backend.
perhaps.
The general idea sounds good. Does adding a few more details help
with understanding, or is it more of an information overload? I'm
thinking of something like:
The "directory rename detection" heuristic in the "merge" and
"interactive" backends can take what happened to paths in the
same directory on the other side of history into account when
deciding whether a new path in that directory should instead be
moved elsewhere. The heuristic produces incorrect results when
the only information available is about files which were changed
on the side of history being rebased, which is why directory
rename detection is disabled when using the "am" backend.
Please let me deposit my objection. This paragraph is not the right
place to explain what directory renme detection is and how it works
under the hood. "works fine" in the original text is the right phrase
here; if there is concern that this induces expectations that cannot be
met, throw in the word "heuristics".
Such as:
Directory rename heuristics work fine in the merge and interactive
backends. It does not in the am backend because...
-- Hannes
