On Thu, Nov 07, 2019 at 09:56:19AM +0700, Doan Tran Cong Danh wrote:

The message file will be used as commit message for the
git-{am,rebase} --continue.

[...]
 	strbuf_addf(&buf, "%s/message", get_dir(opts));
 	if (!file_exists(buf.buf)) {
-		const char *commit_buffer = get_commit_buffer(commit, NULL);
+		const char *encoding = get_commit_output_encoding();
+		const char *commit_buffer = logmsg_reencode(commit, NULL, encoding);

That makes sense, though it's hard to understand the flow of this data
through multiple sequencer invocations. I _think_ this would be fixing a
case like this:

-- >8 --
git init repo
cd repo

# some commits to build off of
echo base >file
git add file
git commit -m base

echo side >file
git add file
git commit -m side

# now make a commit in iso8859-1
git checkout -b side HEAD^
echo iso8859-1 >file
git add file
iconv -f utf8 -t iso8859-1 <<-\EOF |
súbject

bödy
EOF
git -c i18n.commitEncoding=iso8859-1 commit -F -

# and rebase it with the merge strategy, which will fail;
# now .git/rebase-merge/message has iso8859-1 in it
git rebase -m master

# and if we resolve and commit, presumably we'd get a broken commit,
# with iso8859-1 and no encoding header
echo resolved >file
git add file
GIT_EDITOR=: git rebase --continue
-- 8< --

But somehow it all seems to work. The resulting commit has real utf8 in
it. I'm not sure if we pull it from the original commit via "commit -c",
or if it's in one of the other files. But it's not clear to me how
this "message" file is being used.

-Peff