Re^2: Date to be sorted in descending and time in ascending

[aaron_baugher]

I wondered if someone would suggest unpack. I don't use it often enough to remember the format codes without checking the man page, but it seems like it's often the fastest solution for this kind of thing. So I added yours to my benchmark, and found that unpack was slightly slower than substr/substr in this case. (I'd guess that if it were necessary to break the string into three or more pieces, unpack would come out ahead.) Both were still slower than the non-Schwartzian substr/substr sort, as detailed in my other post, though. Results and code:

bannor:~/work/perl/monks$ perl 971240.pl 1000000
          s/iter  stunpack   stsubst plainsort
stunpack    12.4        --      -10%      -31%
stsubst     11.2       11%        --      -24%
plainsort   8.54       46%       31%        --

bannor:~/work/perl/monks$ cat 971240.pl
#!/usr/bin/env perl
use Modern::Perl;
use Benchmark qw(:all);

my @data;
push @data, int(rand(1000000000000))[phone] for (1..$ARGV[0]);

cmpthese( 10, {
        'stunpack'        => \&stunpack,
        'stsubst'         => \&stsubst,
        'plainsort'       => \&plainsort,
});

sub plainsort {
    my @d = sort { 
        substr($b,0,8) <=> substr($a,0,8)
            or substr($a,8) <=> substr($b,8)
    } @data;
}

sub stsubst {
my @d =
    map { $_->[0] }
    sort { $b->[1] <=> $a->[1] or $a->[2] <=> $b->[2] }
    map { [ $_, substr( $_, 0, 8 ), substr( $_, 8)] }
    @data;
}

sub stunpack {
my @d =
    map { $_->[0] }
    sort { $b->[1] <=> $a->[1] or $a->[2] <=> $b->[2] }
    map { [ $_, unpack "a8a6" ] }
    @data;
}

[BrowserUk]

You should add salva's offering to your benchmark.