t6026-merge-attr: child processes must not inherit index.lock handles

On Windows, a file cannot be removed unless all file handles to it have
been released. Hence it is particularly important to close handles when
spawning children (which would probably not even know that they hold on
to those handles).

The example chosen for this test is a custom merge driver that indeed
has no idea that it blocks the deletion of index.lock. The full use case
is a daemon that lives on after the merge, with subsequent invocations
handing off to the daemon, thereby avoiding hefty start-up costs. We
simulate this behavior by simply sleeping one second.

Note that the test only fails on Windows, due to the file locking issue.
Since we have no way to say "expect failure with MINGW, success
otherwise", we simply skip this test on Windows for now.
Signed-off-by: NBen Wijen <ben@wijen.net>
Signed-off-by: NJohannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

t/t6026-merge-attr.sh

@@ -181,4 +181,17 @@ test_expect_success 'up-to-date merge without common ancestor' '
 	)
 '
 
+test_expect_success !MINGW 'custom merge does not lock index' '
+	git reset --hard anchor &&
+	write_script sleep-one-second.sh <<-\EOF &&
+		sleep 1 &
+	EOF
+
+	test_write_lines >.gitattributes \
+		"* merge=ours" "text merge=sleep-one-second" &&
+	test_config merge.ours.driver true &&
+	test_config merge.sleep-one-second.driver ./sleep-one-second.sh &&
+	git merge master
+'
+
 test_done