t3427: add a clarifying comment

dscho · gitster · commit 8c1e24048a0a · 2019-07-31T12:24:06.000-07:00
The flow of this test script is outright confusing, and to start the
endeavor to address that, let's describe what this test is all about,
and how it tries to do it.

Signed-off-by: Johannes Schindelin &lt;johannes.schindelin@gmx.de&gt;
Signed-off-by: Junio C Hamano &lt;gitster@pobox.com&gt;
diff --git a/t/t3427-rebase-subtree.sh b/t/t3427-rebase-subtree.sh
@@ -11,6 +11,34 @@ commit_message() {
 	git log --pretty=format:%s -1 "$1"
 }
 
+# There are a few bugs in the rebase with regards to the subtree strategy, and
+# this test script tries to document them.  First, the following commit history
+# is generated (the onelines are shown, time flows from left to right):
+#
+# master1 - master2 - master3
+#                             \
+# README ---------------------- Add subproject master - master4 - files_subtree/master5
+#
+# Where the merge moves the files master[123].t into the subdirectory
+# files_subtree/ and master4 as well as files_subtree/master5 add files to that
+# directory directly.
+#
+# Then, in subsequent test cases, `git filter-branch` is used to distill just
+# the commits that touch files_subtree/. To give it a final pre-rebase touch,
+# an empty commit is added on top. The pre-rebase commit history looks like
+# this:
+#
+# Add subproject master - master4 - files_subtree/master5 - Empty commit
+#
+# where the root commit adds three files: master1.t, master2.t and master3.t.
+#
+# This commit history is then rebased onto `master3` with the
+# `-Xsubtree=files_subtree` option in three different ways:
+#
+# 1. using `--preserve-merges`
+# 2. using `--preserve-merges` and --keep-empty
+# 3. without specifying a rebase backend
+
 test_expect_success 'setup' '
 	test_commit README &&
 	mkdir files &&
