$/ = "\n-";
open INPUT, "lp.attr" or die "... $!";  # or whatever
while (<INPUT>) {
  chomp;
  my ($name,$attr) = split /\s*:\s*/, $_, 2;
  $name = "-$name" if $. != 1;  # the first record has a -
  my @attrs = split /,\s+/, $attr;
  $CONFIG{$name} = (@attrs == 1) ? $attrs[0] : \@attrs;
}
close INPUT;
[download]

use File::Basename;

opendir(DIR, "/template/dir") or die "can't open dir: $!";
# I'm guessing at the template extension if there is one
@templates = grep { /\.tmpl$/ } readdir DIR;
closedir DIR;

my %printers;

for my $file (@templates) {
 open (FH, $file) or die "Error opening $file: $!";
 my $name;
 while (<FH>) {
  chomp;
  my ($tmp_name, $value) = split /\s*:\s*/;
  $name = $tmp_name if $tmp_name;
  $value =~ s/,$//;
  $printers{basename($file)}{$name}{$value} = undef;
 }
 close FH;
}

# Now %printers contains all valid printer attributes
# and their values
[download]

my (%old, %new);
...
for my $printer (sort keys %old) {
 print("Printer $printer not in new\n"),
  next unless exists $new{$printer};
 for my $name (sort keys %{$old{$printer}}) {
  print("Attribute $name not in new for printer $printer\n"),
   next unless exists $new{$printer}{$name};
  for my $val (sort keys %{$old{$printer}{$name}}) {
   print("Val $val not in new for printer $printer attrib $name\n"),
    next unless exists $new{$printer}{$name}{$val};
  }
 }
}

# Then do the reverse comparision looping over %new
# I'm sure you can figure that out
[download]

It might be stupid, but why not do a diff?

Otherwise I guess this would work:

#!/bin/perl -w
use strict;

my( $file1, $file2)= @ARGV;
my %v1= parse_att( $file1);
my %v2= parse_att( $file2);

$\="\n";
my @new= grep { !$v2{$_} } keys %v1;
my @deleted= grep { !$v1{$_} } keys %v2;

print "new: ", join "\n     ", @new;
print "deleted: ", join "\n         ", @deleted;

sub parse_att
  { my $file= shift;
    my %atts;
    my $field='';
    open( FILE, "<$file") or die "I'd better tell you that the open su
+cceded or I'll get --'ed into oblivion, I'll even report why I failed
+: $!";
    while( <FILE>)
      { chomp;
        my $val;
        if( m/^\s*-([\w-]+)   \s*:\s*([^,]*),?$/)
          { ($field, $val)= ($1, $2); }
        elsif( m/^\s*([^,]*),?$/)
          { $val= $1; }
        else
          { die "uh-oh, looks like this line has a problem: $_"; }
        # why bother with clever data structures
        $atts{"$field $val"}=1;
      }
    return %atts;
  }
[download]

OK... So I tried the above and that was great but in my rush to post. I didn't accurately represent the sample data. So the script is dying on certain unparsable lines. (See below.)

Any advice? Lines like this also exist!

-frame-print-settings-path     : !{dazel-install-directory}!/lib/fm_pr
+int_settings.doc
-font-directories              : !{dazel-install-directory}!/lib/FONTS
+/Soft_Horizons
-ps-init-directories           : !{dazel-install-directory}!/lib/PS
-footer-text-default           : Printed by DAZEL,
                                 Page !{page-number}!,
                                 !{started-printing-time}!
-header-footer-font            : Helvetica-Bold
-header-footer-font-size       : 12
-ps-prologue-file              : ps_prologue.psc,
                                 hp8500dn.ps
-physical-device-type          : ps-printer
-output-document-format-default: ps
-output-bin-xlate              : {face-down, 1},
                                 {top, 1},
                                 {face-up, 2},
                                 {left, 2}
-input-trays-xlate             : {tray1, 0},
                                 {tray2, 2},
                                 {tray3, 3},
                                 {tray4, 4},
                                 {large-capacity, 4}
-ps-printer-echoes-eof         : true
[download]

The problem is probably that some lines include commas, so the [^,]* stops at the first comma instead of the last one

so you need to replace:

if( m/^\s*-([\w-]+)   \s*:\s*([^,]*),?$/)
  { ($field, $val)= ($1, $2); }
elsif( m/^\s*([^,]*),?$/)
  { $val= $1; }
[download]

by

if( m/^\s*-([\w-]+)   \s*:\s*(.*)$/)
  { ($field, $val)= ($1, $2); }
elsif( m/^\s*(.*)?$/)
  { $val= $1; }
$val=~s/,$//;
[download]