Re: Transform Sequence Problem

[hdb]

Using an array of code references and closures:

use strict;
use warnings;

sub fgenerator {
# some code references
        my @f = (
                sub {map {$_ + 1} @_},
                sub {map {log($_)} @_},
                sub {map {$_ * 3} @_},
                );
        my @idx = @_;
        return sub {
                @_ = $f[$_]->(@_) for @idx;
                return @_;
        };
}

# create transform based on indices of function
# read:  first apply f0, then f1, then f2
my $trans1 = fgenerator( 0, 1, 2 );
my $trans2 = fgenerator( 0, 0, 0 );

# some data
my @data = (1, 2, 3);

# call the transforms
my @trans = $trans1->( @data );
print "@trans\n";
@trans = $trans2->( @data );
print "@trans\n";

[hdb]

A few minor changes I like:

• Using a hash instead an array of code refs allows to give names to the transformations.
• A bit of error checking can be added, unknown transformations can be ignored and warned about.
• Instead of passing the indices into the closure, it is more efficient to pass the code references into it. This will generate no additional overhead from the niceties added to the generator function.
It could look like this:

use strict;
use warnings;

sub fgenerator {
        my %f = (
                add_one => sub {map {$_ + 1} @_},
                log     => sub {map {log($_)} @_},
                times_3 => sub {map {$_ * 3} @_},
                );
        my @t = @f{ grep { exists $f{$_} } @_ }; # ignore unknown name
+s
        warn "Don't know sub(s) ".join ", ",(grep { !exists $f{$_} } @
+_) if @t < @_;
        return sub { @_ = $_->(@_) for @t; return @_; };
}

my $trans1 = fgenerator qw( add_one log times_3 );
my $trans2 = fgenerator qw( add_one add_one add_one times_3 fun );

my @data = (1, 2, 3);
my @trans = $trans1->( @data );
print "@trans\n";
@trans = $trans2->( @data );
print "@trans\n";

[hdb]

Something was missing, now I know:

think_of_a_number => sub {map {rand $_} @_},