diff --git a/util/util.go b/util/util.go
index a5ab5c93..fa1d7ce8 100644
--- a/util/util.go
+++ b/util/util.go
@@ -293,6 +293,84 @@ loop:
 	return result
 }
 
+// SafePathClean does path.Clean, but it preserves any trailing slash if it was
+// present in the initial path.
+func SafePathClean(s string) string {
+	hasSlash := strings.HasSuffix(s, "/")
+	clean := path.Clean(s) // removes trailing slashes
+	if hasSlash {          // add it back if it was taken off
+		clean = clean + "/"
+	}
+	return clean
+}
+
+// SegmentedPathSplit splits an absolute path into chunks including their slash.
+// This is similar to the PathSplit function, but the slashes are not lost here!
+// TODO: There is likely a more efficient implementation of this function.
+func SegmentedPathSplit(p string) []string {
+	out := []string{}
+	sponge := ""
+	for i := 0; i < len(p); i++ {
+		sponge = sponge + string(p[i])
+		if string(p[i]) == "/" {
+			out = append(out, sponge) // found one!
+			sponge = ""               // reset
+		}
+	}
+	if sponge != "" {
+		out = append(out, sponge) // the last piece
+	}
+
+	return out
+}
+
+// CommonPathPrefix returns the longest common prefix directory out of all the
+// input paths! This is always a directory and thus ends with a slash, unless
+// all of the paths are identical and are files, or if only one path is given in
+// which case that is returned, or unless no paths are given in which case the
+// empty string is returned. If any of your input paths are not absolute, and as
+// such do not begin with a slash, then the behaviour is undefined.
+func CommonPathPrefix(paths ...string) string {
+	// XXX: I am not a good algorithmist, there is probably a more efficient
+	// way to write this algorithm. Patches are welcome!
+	if len(paths) == 0 {
+		return ""
+	}
+	if len(paths) == 1 {
+		return paths[0]
+	}
+
+	sps := make([][]string, 0)
+	for _, x := range paths {
+		if !strings.HasPrefix(x, "/") {
+			// TODO: panic? error?
+			return "" // undefined behaviour!
+		}
+		//x = SafePathClean(x) // TODO: Should we "safe clean" each path?
+		sp := SegmentedPathSplit(x) // compute once in advance
+		sps = append(sps, sp)
+	}
+
+	z := sps[0] // SegmentedPathSplit(paths[0]) // arbitrarily choose the first one
+	ix := 0
+	ret := ""
+	for {
+		for i := range paths {
+			sp := sps[i]       // SegmentedPathSplit(paths[i])
+			if len(sp) <= ix { // one path is longer
+				return ret
+			}
+
+			if sp[ix] != z[ix] { // the chunk differs!
+				return ret
+			}
+		}
+
+		ret = ret + z[ix] // append what we've got so far!
+		ix++
+	}
+}
+
 // PathPrefixDelta returns the delta of the path prefix, which tells you how
 // many path tokens different the prefix is.
 func PathPrefixDelta(p, prefix string) int {
diff --git a/util/util_test.go b/util/util_test.go
index e88c2ed1..a06006ec 100644
--- a/util/util_test.go
+++ b/util/util_test.go
@@ -795,6 +795,117 @@ func TestUtilT11(t *testing.T) {
 	}
 }
 
+func TestSegmentedPathSplit(t *testing.T) {
+	if ex, out := []string{}, SegmentedPathSplit(
+		"",
+	); !reflect.DeepEqual(out, ex) {
+		t.Errorf("expected: %v got: %v", ex, out)
+	}
+
+	if ex, out := []string{"/"}, SegmentedPathSplit(
+		"/",
+	); !reflect.DeepEqual(out, ex) {
+		t.Errorf("expected: %v got: %v", ex, out)
+	}
+
+	if ex, out := []string{"/", "foo/", "bar/"}, SegmentedPathSplit(
+		"/foo/bar/",
+	); !reflect.DeepEqual(out, ex) {
+		t.Errorf("expected: %v got: %v", ex, out)
+	}
+
+	if ex, out := []string{"/", "foo/", "bar"}, SegmentedPathSplit(
+		"/foo/bar",
+	); !reflect.DeepEqual(out, ex) {
+		t.Errorf("expected: %v got: %v", ex, out)
+	}
+}
+
+func TestCommonPathPrefix1(t *testing.T) {
+	if ex, out := "/foo/whatever2/", CommonPathPrefix(
+		"/foo/whatever2/",
+		"/foo/whatever2/",
+		"/foo/whatever2/",
+	); out != ex {
+		t.Errorf("expected: %v got: %v", ex, out)
+	}
+}
+
+func TestCommonPathPrefix2(t *testing.T) {
+	if ex, out := "/whatever1", CommonPathPrefix(
+		"/whatever1",
+	); out != ex {
+		t.Errorf("expected: %v got: %v", ex, out)
+	}
+	if ex, out := "/whatever2", CommonPathPrefix(
+		"/whatever2",
+		"/whatever2",
+	); out != ex {
+		t.Errorf("expected: %v got: %v", ex, out)
+	}
+	if ex, out := "/foo/whatever1", CommonPathPrefix(
+		"/foo/whatever1",
+		"/foo/whatever1",
+	); out != ex {
+		t.Errorf("expected: %v got: %v", ex, out)
+	}
+	if ex, out := "/foo/whatever2", CommonPathPrefix(
+		"/foo/whatever2",
+		"/foo/whatever2",
+		"/foo/whatever2",
+	); out != ex {
+		t.Errorf("expected: %v got: %v", ex, out)
+	}
+
+	if ex, out := "/whatever3/", CommonPathPrefix(
+		"/whatever3/",
+		"/whatever3/",
+	); out != ex {
+		t.Errorf("expected: %v got: %v", ex, out)
+	}
+	if ex, out := "/foo/whatever3/", CommonPathPrefix(
+		"/foo/whatever3/",
+		"/foo/whatever3/",
+	); out != ex {
+		t.Errorf("expected: %v got: %v", ex, out)
+	}
+	if ex, out := "/foo/whatever4/", CommonPathPrefix(
+		"/foo/whatever4/",
+		"/foo/whatever4/",
+		"/foo/whatever4/",
+	); out != ex {
+		t.Errorf("expected: %v got: %v", ex, out)
+	}
+
+	if ex, out := "/", CommonPathPrefix(
+		"/foo/bar",
+		"/bar/baz/",
+		"/baz/bing/wow",
+	); out != ex {
+		t.Errorf("expected: %v got: %v", ex, out)
+	}
+
+	if ex, out := "/foo/", CommonPathPrefix(
+		"/foo/bar/",
+		"/foo/bar/dude",
+		"/foo/bar", // this is not the same as /foo/bar/ !
+	); out != ex {
+		t.Errorf("expected: %v got: %v", ex, out)
+	}
+
+	// If we want to "safe clean" each path, then this test should be added.
+	//if ex, out := "/home/james/tmp/", CommonPathPrefix(
+	//	"/home/james/tmp/coverage/test",
+	//	"/home/james/tmp/covert/operator",
+	//	"/home/james/tmp/coven/members",
+	//	"/home//james/tmp/coventry",
+	//	"/home/james/././tmp/covertly/foo",
+	//	"/home/luser/../james/tmp/coved/bar",
+	//); out != ex {
+	//	t.Errorf("expected: %v got: %v", ex, out)
+	//}
+}
+
 func TestUtilFlattenListWithSplit1(t *testing.T) {
 	{
 		in := []string{} // input