From: Zygo Blaxell <zblaxell@waya.furryterror.org>
Date: Sat, 9 Jan 2010 01:51:45 +0000 (-0500)
Subject: Merge branch 'performance'
X-Git-Tag: dm6-0.20100514~31
X-Git-Url: http://git.hungrycats.org/cgi-bin/gitweb.cgi?p=dupemerge;a=commitdiff_plain;h=776aeabce8add0b31e16bb91b13e6cb9fcee682a;hp=dc1f8376951d597b2b78191254b9ec493eaf16ff

Merge branch 'performance'

Conflicts:
	faster-dupemerge
---

diff --git a/faster-dupemerge b/faster-dupemerge
index 0b43a3c..d35f591 100755
--- a/faster-dupemerge
+++ b/faster-dupemerge
@@ -238,7 +238,7 @@ my @find_command = ('find', @directories, @extra_find_opts, '-type', 'f');
 my $printf_string = '%s ' .
 	($collapse_access    ? '0 0 0 ' : '%U %G %m ') .
 	($collapse_timestamp ? '0 '     : '%T@ ') .
-	'%i %p\0';
+	'%D:%i %p\0';
 
 push(@find_command, '!', '-empty') unless $collapse_zero;
 push(@find_command, '-printf', $printf_string);
@@ -305,14 +305,14 @@ sub merge_files {
 	}
 
 	print STDERR "Merging...\n" if $debug;
-	foreach my $candidate (@candidate_list) {
+	foreach my $candidate (sort @candidate_list) {
 		print STDERR "\tDigesting candidate $candidate\n" if $debug;
 		my $ok = 0;
 		my $digest;
 
 hash_file:
 
-		foreach my $filename (keys(%{$inode_to_file_name{$candidate}})) {
+		foreach my $filename (sort keys(%{$inode_to_file_name{$candidate}})) {
 			print STDERR "\t\tDigesting file $filename\n" if $debug;
 			if ((-l $filename) || ! -f _) {
 				warn "Bogon file " . tick_quote($filename);
@@ -343,8 +343,8 @@ hash_file:
 link_start:
 
 				until ($finished) {
-					my @incumbent_names = keys(%{$inode_to_file_name{$incumbent}});
-					my @candidate_names = keys(%{$inode_to_file_name{$candidate}});
+					my @incumbent_names = sort keys(%{$inode_to_file_name{$incumbent}});
+					my @candidate_names = sort keys(%{$inode_to_file_name{$candidate}});
 					print STDERR "\t\tLinks to $incumbent:", join("\n\t\t\t", '', @incumbent_names), "\n" if $debug;
 					print STDERR "\t\tLinks to $candidate:", join("\n\t\t\t", '', @candidate_names), "\n" if $debug;
 
@@ -443,6 +443,12 @@ candidate_file:
 									my $link_done = 0;
 
 									my ($from_file, $to_file, $from_inode, $to_inode, $from_nlink, $to_nlink);
+
+									# If the candidate has more links than incumbent, replace incumbent with candidate.
+									# If the incumbent has more links than candidate, replace candidate with incumbent.
+									# If the link counts are equal, we saw incumbent first, so keep the incumbent.
+									# "We saw incumbent first" is significant because we explicitly sort the inodes.
+									# Thank Johannes Niess for this idea.
 									if ($candidate_nlink > $incumbent_nlink) {
 										$from_file = $candidate_file;
 										$to_file = $incumbent_file;
@@ -537,7 +543,7 @@ end_merge:
 }
 
 while (<FIND>) {
-	my ($weak_key, $inode, $name) = m/^(\d+ \d+ \d+ \d+ -?[\d.]+) (\d+) (.+)\0$/so;
+	my ($weak_key, $inode, $name) = m/^(\d+ \d+ \d+ \d+ -?[\d.]+) (\d+:\d+) (.+)\0$/so;
 	die "read error: $!\nLast input line was '$_'" unless defined($name);
 
 	print STDERR "weak_key=$weak_key inode=$inode name=$name\n" if $debug;