| Hash::FieldHash(3pm) | User Contributed Perl Documentation | Hash::FieldHash(3pm) |
Hash::FieldHash - Lightweight field hash for inside-out objects
This document describes Hash::FieldHash version 0.15.
use Hash::FieldHash qw(:all);
fieldhash my %foo;
fieldhashes \my(%bar, %baz);
{
my $o = Something->new();
$foo{$o} = 42;
print $foo{$o}; # => 42
}
# when $o is released, $foo{$o} is also deleted,
# so %foo is empty in here.
# in a class
{
package Foo;
use Hash::FieldHash qw(:all);
fieldhash my %bar, 'bar'; # make an accessor
}
my $obj = bless {}, 'Foo';
$obj->bar(10); # does $bar{$obj} = 10
"Hash::FieldHash" provides the field hash mechanism which supports the inside-out technique.
You may know "Hash::Util::FieldHash". It's a very useful module, but too complex to understand the functionality and only available in 5.10. "H::U::F::Compat" is available for pre-5.10, but it is too slow to use.
This is a better alternative to "H::U::F" with following features:
Optional $name and $package indicate the name of the field, which will create rw-accessors, using the same name as $name.
Returns nothing.
Returns nothing.
Returns $object.
If the "-fully_qualify" option is supplied , field keys are fully qualified.
For example:
package MyClass;
use FieldHash qw(:all);
fieldhash my %foo => 'foo';
sub new{
my $class = shift;
my $self = bless {}, $class;
return from_hash($self, @_);
}
package MyDerivedClass;
use parent -norequire => 'MyClass';
use FieldHash qw(:all);
fieldhash my %bar => 'bar';
package main;
my $o = MyDerivedClass->new(foo => 10, bar => 20);
my $p = MyDerivedClass->new('MyClass::foo' => 10, 'MyDerivedClass::bar' => 20);
use Data::Dumper;
print Dumper($o->to_hash());
# $VAR1 = { foo => 10, bar => 20 }
print Dumper($o->to_hash(-fully_qualify));
# $VAR1 = { 'MyClass::foo' => 10, 'MyDerived::bar' => 20 }
As "Hash::Util::FieldHash" does, "Hash::FieldHash" fully supports threading using the "CLONE" method.
"Hash::FieldHash" itself does not leak memory, but it may leak memory when you uses hash references as field hash keys because of an issue of perl 5.10.0.
"Hash::FieldHash" accepts only references and registered addresses as its keys, whereas "Hash::Util::FieldHash" accepts any type of scalars.
According to "The Generic Object" in Hash::Util::FieldHash, Non-reference keys in "H::U::F" are used for class fields. That is, all the fields defined by "H::U::F" act as both object fields and class fields by default. It seems confusing; if you do not want them to be class fields, you must check the type of $self explicitly. In addition, these class fields are never inherited. This behavior seems problematic, so "Hash::FieldHash" restricts the type of keys.
While "Hash::Util::FieldHash" uses "refaddr" as the IDs of field hash keys, "Hash::FieldHash" allocates arbitrary integers as the IDs.
The accessors "fieldhash()" creates are chainable accessors. That is, it returns the $object (i.e. $self) with a parameter, where as it returns the $value without it.
For example:
my $o = YourClass->new();
$o->foo(42); # returns $o itself
my $value = $o->foo(); # retuns 42
Perl 5.8.5 or later, and a C compiler.
No bugs have been reported.
Please report any bugs or feature requests to the author.
Hash::Util::FieldHash.
Hash::Util::FieldHash::Compat.
"Magic Virtual Tables" in perlguts.
Class::Std describes the inside-out technique.
Fuji, Goro (gfx) <gfuji(at)cpan.org>.
Copyright (c) 2009-2010, Fuji, Goro. All rights reserved.
This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself.
| 2018-11-02 | perl v5.28.0 |