NAME
accessors - create accessor methods in caller's package.
SYNOPSIS
package Foo;
use accessors qw( foo bar baz );
my $obj = bless {}, 'Foo';
# generates chaining accessors
# that you can set like this:
$obj->foo( 'hello ' )
->bar( 'world' )
->baz( "!\n" );
# you get the values by passing no params:
print $obj->foo, $obj->bar, $obj->baz;
DESCRIPTION
The accessors pragma lets you create simple accessors at compile-time.
This saves you from writing them by hand, which tends to result in
*cut-n-paste* errors and a mess of duplicated code. It can also help you
reduce the ammount of unwanted *direct-variable access* that may creep
into your codebase when you're feeling lazy. accessors was designed with
laziness in mind.
Method-chaining accessors are generated by default. Note that you can
still use accessors::chained directly for reasons of backwards
compatability.
See accessors::classic for accessors that always return the current
value if you don't like method chaining.
GENERATED METHODS
accessors will generate methods that return the current object on set:
sub foo {
my $self = shift;
if (@_) { $self->{-foo} = shift; return $self; }
else { return $self->{-foo}; }
}
This way they can be *chained* together.
Why prepend the dash?
The dash ("-") is prepended to the property name for a few reasons:
* interoperability with Error.
* to make it difficult to accidentally access the property directly
ala:
use accessors qw( foo );
$obj->{foo}; # prevents this by mistake
$obj->foo; # when you probably meant this
(this might sound woolly, but it's easy enough to do).
* syntactic sugar (this *is* woolly :).
You shouldn't care too much about how the property is stored anyway - if
you do, you're likely trying to do something special (and should really
consider writing the accessors out long hand), or it's simply a matter
of preference in which case you can use accessors::classic, or sub-class
this module.
PERFORMANCE
There is little-to-no performace hit when using generated accessors; in
fact there is usually a performance gain.
* typically *10-30% faster* than hard-coded accessors (like the above
example).
* typically *1-15% slower* than *optimized* accessors (less readable).
* typically a *small* performance hit at startup (accessors are
created at compile-time).
* uses the same anonymous sub to reduce memory consumption (sometimes
by 80%).
See the benchmark tests included with this distribution for more
details.
MOTIVATION
The main difference between the accessors pragma and other accessor
generators is simplicity.
* interface
use accessors qw( ... ) is as easy as it gets.
* a pragma
it fits in nicely with the base pragma:
use base qw( Some::Class );
use accessors qw( foo bar baz );
and accessors get created at compile-time.
* no bells and whistles
The module is extensible instead.
SUB-CLASSING
If you prefer a different style of accessor or you need to do something
more complicated, there's nothing to stop you from sub-classing. It
should be pretty easy. Look through accessors::classic, accessors::ro,
and accessors::rw to see how it's done.
CAVEATS
Classes using blessed scalarrefs, arrayrefs, etc. are not supported for
sake of simplicity. Only hashrefs are supported.
THANKS
Thanks to Michael G. Schwern for indirectly inspiring this module, and
for his feedback & suggestions.
Also to Paul Makepeace and David Wright for showing me faster accessors,
to chocolateboy for his contributions, the CPAN Testers for their bug
reports, and to James Duncan and people on London.pm for their feedback.
AUTHOR
Steve Purkis <
[email protected]>
SEE ALSO
accessors::classic, accessors::chained
Similar and related modules:
base, fields, Class::Accessor, Class::Struct, Class::Methodmaker,
Class::Generate, Class::Class, Class::Tangram, Object::Tiny
