diff --git a/contrib/diff-highlight/README b/contrib/diff-highlight/README
index 1b7b6df8ebbdfccab8a632d883c321b4ee2919a4..502e03b3058e947835d396540af39c83a45d42aa 100644
--- a/contrib/diff-highlight/README
+++ b/contrib/diff-highlight/README
@@ -14,13 +14,15 @@ Instead, this script post-processes the line-oriented diff, finds pairs
 of lines, and highlights the differing segments.  It's currently very
 simple and stupid about doing these tasks. In particular:
 
-  1. It will only highlight a pair of lines if they are the only two
-     lines in a hunk.  It could instead try to match up "before" and
-     "after" lines for a given hunk into pairs of similar lines.
-     However, this may end up visually distracting, as the paired
-     lines would have other highlighted lines in between them. And in
-     practice, the lines which most need attention called to their
-     small, hard-to-see changes are touching only a single line.
+  1. It will only highlight hunks in which the number of removed and
+     added lines is the same, and it will pair lines within the hunk by
+     position (so the first removed line is compared to the first added
+     line, and so forth). This is simple and tends to work well in
+     practice. More complex changes don't highlight well, so we tend to
+     exclude them due to the "same number of removed and added lines"
+     restriction. Or even if we do try to highlight them, they end up
+     not highlighting because of our "don't highlight if the whole line
+     would be highlighted" rule.
 
   2. It will find the common prefix and suffix of two lines, and
      consider everything in the middle to be "different". It could
@@ -55,3 +57,96 @@ following in your git configuration:
 	show = diff-highlight | less
 	diff = diff-highlight | less
 ---------------------------------------------
+
+Bugs
+----
+
+Because diff-highlight relies on heuristics to guess which parts of
+changes are important, there are some cases where the highlighting is
+more distracting than useful. Fortunately, these cases are rare in
+practice, and when they do occur, the worst case is simply a little
+extra highlighting. This section documents some cases known to be
+sub-optimal, in case somebody feels like working on improving the
+heuristics.
+
+1. Two changes on the same line get highlighted in a blob. For example,
+   highlighting:
+
+----------------------------------------------
+-foo(buf, size);
++foo(obj->buf, obj->size);
+----------------------------------------------
+
+   yields (where the inside of "+{}" would be highlighted):
+
+----------------------------------------------
+-foo(buf, size);
++foo(+{obj->buf, obj->}size);
+----------------------------------------------
+
+   whereas a more semantically meaningful output would be:
+
+----------------------------------------------
+-foo(buf, size);
++foo(+{obj->}buf, +{obj->}size);
+----------------------------------------------
+
+   Note that doing this right would probably involve a set of
+   content-specific boundary patterns, similar to word-diff. Otherwise
+   you get junk like:
+
+-----------------------------------------------------
+-this line has some -{i}nt-{ere}sti-{ng} text on it
++this line has some +{fa}nt+{a}sti+{c} text on it
+-----------------------------------------------------
+
+   which is less readable than the current output.
+
+2. The multi-line matching assumes that lines in the pre- and post-image
+   match by position. This is often the case, but can be fooled when a
+   line is removed from the top and a new one added at the bottom (or
+   vice versa). Unless the lines in the middle are also changed, diffs
+   will show this as two hunks, and it will not get highlighted at all
+   (which is good). But if the lines in the middle are changed, the
+   highlighting can be misleading. Here's a pathological case:
+
+-----------------------------------------------------
+-one
+-two
+-three
+-four
++two 2
++three 3
++four 4
++five 5
+-----------------------------------------------------
+
+   which gets highlighted as:
+
+-----------------------------------------------------
+-one
+-t-{wo}
+-three
+-f-{our}
++two 2
++t+{hree 3}
++four 4
++f+{ive 5}
+-----------------------------------------------------
+
+   because it matches "two" to "three 3", and so forth. It would be
+   nicer as:
+
+-----------------------------------------------------
+-one
+-two
+-three
+-four
++two +{2}
++three +{3}
++four +{4}
++five 5
+-----------------------------------------------------
+
+   which would probably involve pre-matching the lines into pairs
+   according to some heuristic.
diff --git a/contrib/diff-highlight/diff-highlight b/contrib/diff-highlight/diff-highlight
index d8938982e413a9bf994bd12386121249c888649d..c4404d49c9968608510809309b26e2d08eec8810 100755
--- a/contrib/diff-highlight/diff-highlight
+++ b/contrib/diff-highlight/diff-highlight
@@ -1,28 +1,37 @@
 #!/usr/bin/perl
 
+use warnings FATAL => 'all';
+use strict;
+
 # Highlight by reversing foreground and background. You could do
 # other things like bold or underline if you prefer.
 my $HIGHLIGHT   = "\x1b[7m";
 my $UNHIGHLIGHT = "\x1b[27m";
 my $COLOR = qr/\x1b\[[0-9;]*m/;
+my $BORING = qr/$COLOR|\s/;
 
-my @window;
+my @removed;
+my @added;
+my $in_hunk;
 
 while (<>) {
-	# We highlight only single-line changes, so we need
-	# a 4-line window to make a decision on whether
-	# to highlight.
-	push @window, $_;
-	next if @window < 4;
-	if ($window[0] =~ /^$COLOR*(\@| )/ &&
-	    $window[1] =~ /^$COLOR*-/ &&
-	    $window[2] =~ /^$COLOR*\+/ &&
-	    $window[3] !~ /^$COLOR*\+/) {
-		print shift @window;
-		show_pair(shift @window, shift @window);
+	if (!$in_hunk) {
+		print;
+		$in_hunk = /^$COLOR*\@/;
+	}
+	elsif (/^$COLOR*-/) {
+		push @removed, $_;
+	}
+	elsif (/^$COLOR*\+/) {
+		push @added, $_;
 	}
 	else {
-		print shift @window;
+		show_hunk(\@removed, \@added);
+		@removed = ();
+		@added = ();
+
+		print;
+		$in_hunk = /^$COLOR*[\@ ]/;
 	}
 
 	# Most of the time there is enough output to keep things streaming,
@@ -38,23 +47,40 @@ while (<>) {
 	}
 }
 
-# Special case a single-line hunk at the end of file.
-if (@window == 3 &&
-    $window[0] =~ /^$COLOR*(\@| )/ &&
-    $window[1] =~ /^$COLOR*-/ &&
-    $window[2] =~ /^$COLOR*\+/) {
-	print shift @window;
-	show_pair(shift @window, shift @window);
-}
-
-# And then flush any remaining lines.
-while (@window) {
-	print shift @window;
-}
+# Flush any queued hunk (this can happen when there is no trailing context in
+# the final diff of the input).
+show_hunk(\@removed, \@added);
 
 exit 0;
 
-sub show_pair {
+sub show_hunk {
+	my ($a, $b) = @_;
+
+	# If one side is empty, then there is nothing to compare or highlight.
+	if (!@$a || !@$b) {
+		print @$a, @$b;
+		return;
+	}
+
+	# If we have mismatched numbers of lines on each side, we could try to
+	# be clever and match up similar lines. But for now we are simple and
+	# stupid, and only handle multi-line hunks that remove and add the same
+	# number of lines.
+	if (@$a != @$b) {
+		print @$a, @$b;
+		return;
+	}
+
+	my @queue;
+	for (my $i = 0; $i < @$a; $i++) {
+		my ($rm, $add) = highlight_pair($a->[$i], $b->[$i]);
+		print $rm;
+		push @queue, $add;
+	}
+	print @queue;
+}
+
+sub highlight_pair {
 	my @a = split_line(shift);
 	my @b = split_line(shift);
 
@@ -101,8 +127,14 @@ sub show_pair {
 		}
 	}
 
-	print highlight(\@a, $pa, $sa);
-	print highlight(\@b, $pb, $sb);
+	if (is_pair_interesting(\@a, $pa, $sa, \@b, $pb, $sb)) {
+		return highlight_line(\@a, $pa, $sa),
+		       highlight_line(\@b, $pb, $sb);
+	}
+	else {
+		return join('', @a),
+		       join('', @b);
+	}
 }
 
 sub split_line {
@@ -111,7 +143,7 @@ sub split_line {
 	       split /($COLOR*)/;
 }
 
-sub highlight {
+sub highlight_line {
 	my ($line, $prefix, $suffix) = @_;
 
 	return join('',
@@ -122,3 +154,20 @@ sub highlight {
 		@{$line}[($suffix+1)..$#$line]
 	);
 }
+
+# Pairs are interesting to highlight only if we are going to end up
+# highlighting a subset (i.e., not the whole line). Otherwise, the highlighting
+# is just useless noise. We can detect this by finding either a matching prefix
+# or suffix (disregarding boring bits like whitespace and colorization).
+sub is_pair_interesting {
+	my ($a, $pa, $sa, $b, $pb, $sb) = @_;
+	my $prefix_a = join('', @$a[0..($pa-1)]);
+	my $prefix_b = join('', @$b[0..($pb-1)]);
+	my $suffix_a = join('', @$a[($sa+1)..$#$a]);
+	my $suffix_b = join('', @$b[($sb+1)..$#$b]);
+
+	return $prefix_a !~ /^$COLOR*-$BORING*$/ ||
+	       $prefix_b !~ /^$COLOR*\+$BORING*$/ ||
+	       $suffix_a !~ /^$BORING*$/ ||
+	       $suffix_b !~ /^$BORING*$/;
+}