octopus: do not do AND'ed merge base.

When doing an octopus, we incorrectly used the previous merge
base as the reference to compute next merge base.  This was
unnecessary, because that can never be better than using the
original HEAD.  And that is far simpler as well ;-).

Signed-off-by: Junio C Hamano <junkio@cox.net>

1 files changed, 20 insertions, 5 deletions

diff --git a/git-merge-octopus.sh b/git-merge-octopus.sh
index aa1cd2f106..bb58e22a18 100755
--- a/git-merge-octopus.sh
+++ b/git-merge-octopus.sh
@@ -5,6 +5,9 @@
 # Resolve two or more trees.
 #
 
+LF='
+'
+
 # The first parameters up to -- are merge bases; the rest are heads.
 bases= head= remotes= sep_seen=
 for arg
@@ -42,14 +45,18 @@ CNT=1 ;# counting our head
 NON_FF_MERGE=0
 for SHA1 in $remotes
 do
-	common=$(git-merge-base $MRC $SHA1) ||
+	common=$(git-merge-base --all $MRC $SHA1) ||
 		die "Unable to find common commit with $SHA1"
 
-	if test "$common" = $SHA1
-	then
+	case "$common" in
+	?*"$LF"?*)
+		die "Not trivially mergeable."
+		;;
+	$SHA1)
 		echo "Already up-to-date with $SHA1"
 		continue
-	fi
+		;;
+	esac
 
 	CNT=`expr $CNT + 1`
 	PARENT="$PARENT -p $SHA1"
@@ -79,7 +86,15 @@ do
 		exit 2 ; # Automatic merge failed; should not be doing Octopus
 		next=$(git-write-tree 2>/dev/null)
 	fi
-	MRC=$common
+
+	# We have merged the other branch successfully.  Ideally
+	# we could implement OR'ed heads in merge-base, and keep
+	# a list of commits we have merged so far in MRC to feed
+	# them to merge-base, but we approximate it by keep using
+	# the current MRC.  We used to update it to $common, which
+	# was incorrectly doing AND'ed merge-base here, which was
+	# unneeded.
+
 	MRT=$next
 done