in reply to Fast Building of SQL Statements

This method is fast, as the following benchmark shows:

Yeah, but it's also wrong, as the following code shows: 

#!/usr/bin/perl

use strict;
use warnings;

use Benchmark qw( cmpthese );

my $count = shift || 100000;
my @test_array = (0 .. 99);
my (@array1, @array2);

sub map_bench
{
    @array1 = map { '?' } @test_array;
}

sub x_bench
{
    @array2 = '?' x @test_array;
}

map_bench;
x_bench;

print "\@test_array has ", scalar @test_array, " elements\n";
print "\@array1 has ", scalar @array1, " elements\n";
print "\@array2 has ", scalar @array2, " elements\n";

__END__
@test_array has 100 elements
@array1 has 100 elements
@array2 has 1 elements

[download]

The x operator will only return a multi-element list if all of the following conditions are met:

And the last condition isn't met. Here's a better benchmark, x is still winning, but not with such a margin: 
#!/usr/bin/perl

use strict;
use warnings;

use Benchmark qw /cmpthese/;

our @test_array = (0 .. 99);
our (@array1, @array2);

cmpthese -2 => {
    map => '@::array1 = map {"?"} @::test_array',
    x   => '@::array2 = ("?") x @::test_array'
};

die "Benchmark failed" unless @test_array == @array1 &&
                              @test_array == @array2;

__END__

Benchmark: running map, x for at least 2 CPU seconds...
       map:  3 wallclock secs ( 2.09 usr +  0.03 sys =  2.12 CPU) @ 46
+08.49/s (n=9770)
         x:  2 wallclock secs ( 2.28 usr +  0.00 sys =  2.28 CPU) @ 14
+508.33/s (n=33079)
       Rate  map    x
map  4608/s   -- -68%
x   14508/s 215%   --

[download]

Abigail

Replies are listed 'Best First'.
Re: Re: Fast Building of SQL Statements
by zby (Vicar) on Aug 28, 2003 at 07:57 UTC
    To defend his technique I slightly modified his benchmarking script: 
    #!/usr/bin/perl
use strict;
use warnings;
use Benchmark qw( cmpthese );

my $count = shift || 100000;
my @test_array = (0 .. 99);

sub map_bench 
{
    my $result = join(', ', map { '?' } @test_array);
}

sub x_bench 
{
    my $result = '?, ' x @test_array;
}

cmpthese($count, {
    map => \&map_bench, 
    x   => \&x_bench, 
});

__END__

Benchmark: timing 100000 iterations of map, x...
       map: 28 wallclock secs (27.94 usr +  0.00 sys = 27.94 CPU) @ 35
+78.97/s (n=100000)
         x:  1 wallclock secs ( 0.77 usr +  0.00 sys =  0.77 CPU) @ 12
+9533.68/s (n=100000)
        Rate   map     x
map   3579/s    --  -97%
x   129534/s 3519%    --

    [download]
    Update: Unfortunately I forgot about the first reason of using join - getting rid of the last comma. Perhaps it should be 
    my $result = ('?, ' x $#test_array) . '?'

    [download]
    It's not as elegant as it was. But still simpler than 
    my $result = join(', ', map { '?' } @test_array);

    [download]
[reply]
[d/l]
[select]

In Section Meditations