There's also strftime in POSIX.

$ perl -MPOSIX=strftime -le'print strftime "%a %b %e %Y", localtime'
Thu Dec 17 2009
[download]

You could also use substr to remove the unwanted parts.

$ perl -le'substr(my $now = localtime, 11, 9, ""); print $now'
Thu Dec 17 2009
[download]

Benchmark results

             Rate   splice strftime    slice    subst   substr
splice    93974/s       --     -23%     -40%     -49%     -64%
strftime 121619/s      29%       --     -23%     -33%     -53%
slice    157894/s      68%      30%       --     -14%     -40%
subst    182658/s      94%      50%      16%       --     -30%
substr   261283/s     178%     115%      65%      43%       --
[download]

It surprises me that strftime is slower than slice and subst.

There's no surprise that substr is the fastest, but I don't know if it's safe. strftime is of course the safest.

Honestly, though. I don't see how the difference in speed observed could possibly matter in practice.

Benchmark code:

use strict;
use warnings;

use Benchmark qw( cmpthese );
use POSIX     qw( strftime );

sub using_splice {
    my @t = split /\s/, $_[0];
    splice @t, 3, 1;
    return join ' ', @t;
}

sub using_slice {
    return join " ", ( split " ", $_[0] )[0,1,2,4];
}

my %tests = (
    splice   => 'my $now = using_splice(scalar(localtime));',
    slice    => 'my $now = using_slice(scalar(localtime));',
    subst    => '( my $now = localtime ) =~ s/(?:\d\d:){2}\d\d //;',
    strftime => 'my $now = strftime("%a %b %d %Y", localtime);',
    substr   => 'substr(my $now = localtime, 11, 9, "");',
);

$_ = "use strict; use warnings; $_"
    for values %tests;

cmpthese(-3, \%tests);
[download]

Update: Added slice. Improved subst.