Personally, I would use GNU diff (available for most platforms) to do the quickest possible comparison for me...

diff --recursive --brief dir1 dir2
[download]

If I still felt it necessary to get Perl involved (say, for scheduled automation), I would capture and parse the results as appropriate.

This is a perfect example when to not use File::Find. It really isn't very hard to roll your own directory tree searcher while the File::Find call-back interface makes it impossible to traverse two directory trees at once.

Beside the standard gotchas to watch out for (don't follow symbolic links unless you do the extra work required, note that readdir of other than "." means you have to prepend the directory before you use the returned file names to get information about the files, don't use a global directory handle in opendir calls of a recursive subroutine, don't recurse into "." nor "..") also realize that readdir doesn't return file names in sorted order (while almost all globs do) so you'll want to sort (and ignore case when you sort if dealing with a file system that ignores case) before doing a merge-sort comparison to find missing/added/different files/directories (or files that became directories or vice versa).

-- Randal L. Schwartz, Perl hacker
Be sure to read my standard disclaimer if this is a reply.

Here's a little idea for you. If you are just checking modified times and not actual content, you can simple do something like:

1. Do a File::Find on the old directory structure
2. Store the information in a hash with the filenames(with full paths) as keys and the modified date/time as the value.
3. Once you finish, do a File::Find on the new directory structure
4. Compare the modify date/times of the new files with what you have stored for the old files

I don't know how many files you are talking about? Thousands? Millions? Don't underestimate Perl's speed, if all you are checking is modify times then your bottleneck will be the Disk I/O not your script. And that would be the case with other solutions as well. If you are concerned about time and efficiency, you can separate the two loops above and have the resulting hash from the File::Find on the old directory structure stored using the Storable module. That way you can cache the mod times of the files before you need to compare them to the new directory structure and then just retrieve them when you run the script against the new directory.

HTH

use Tie::TextDir;
use Digest::MD5;
my ($dirA, $dirB) = @ARGV;
tie my %filsA, 'Tie::TextDir', ($dirA || "."), 'rw';
tie my %filsB, 'Tie::TextDir', ($dirB || "."), 'rw';
my (%common, @uniqA, @uniqB, @modified);
map { 
    $common{$_} = Digest::MD5::md5_base64($filsA{$_}) if exists $filsB
+{$_} 
} keys %filsA;
map { push (@uniqA, $_) if ! exists $filsB{$_} } keys %filsA;
map { push (@uniqB, $_) if ! exists $filsA{$_} } keys %filsB;
@modified = grep { 
    ! (Digest::MD5::md5_base64($filsB{$_}) eq $common{$_} )
    } keys %common;
untie %filsA; untie %filsB;
print "** common to $dirA, $dirB **\n" . join "\n", keys %common;
print "\n** unique to $dirA **\n" . join "\n", @uniqA;
print "\n** unique to $dirB **\n" . join "\n", @uniqA;
print "\n** modified files  **\n" . join "\n", @modified;
print "\n** finito **\n";
[download]

** common to chaffwin, chaffwin2 **
TO_DO
md5.h
README
COPYING
ChangeLog
Makefile
chaffwin.el
chaffwin.o
md5.o
chaffwin.c
chaffwin.pl
md5.c
CVS
** unique to chaffwin **
chaffwin.exe
oot
** unique to chaffwin2 **
chaffwin.exe
oot
** modified files  **
chaffwin.o
** finito **
[download]

Bye
 PetaMem

SYNOPSIS use File::Dircmp; @r = dircmp($dir1, $dir2, $diff, $suppress);

DESCRIPTION The dircmp command examines dir1 and dir2 and generates various tabulated information about the contents of the directories. Listings of files that are unique to each directory are generated for all the options. If no option is entered, a list is output indicating whether the file names common to both directories have the same contents.