NAME

Object::POOF - Perl Object-Oriented Framework

SYNOPSIS

 package MyApp::Model;
 require Object::POOF;
 @ISA = qw( Object::POOF );
 
 sub init {
    my $self = shift;
    $self->SUPER::init;                  # call parent init
    # other custom setup...
    return 1;
 }
 
 sub some_method {
    my ($self,$p) = @_;
    my $var1 = $p->{var1};
    my $var2 = $p->{var2};
    # do something....
    return;
 }
 
 package main;
 require MyApp::Model;
 require MyApp::DB;  # see Object::POOF::DB;
 my $model = MyApp::Model->new(   id      	=> $id;
                                  nondbprop => "nondbval",
                                  db     	=> $db   );
 print $model->id."\n";
 print $model->data->{quick_dbfield1}."\n";
 print $model->{data}->{quick_dbfield2}."\n";    # cached data values
 print $model->data_info->{info_dbfield1}."\n";  # suppl. info data
 $model->destroy;
 
 $model = MyApp::Model->new(	db		=> $db,
 								where	=>	{ quick_dbfield1 => $val1,
											  info_dbfield1  => $val2
											}
							);

 $model->{save}->{quick_dbfield1}       = "new_value1";
 $model->{save_info}->{info_dbfield1}   = "new_value2";
 $model->save;
 $model->commit;
 $model->data;   # re-select and re-cache new data values
 
 # MyApp::DB object can be passed around as a shared transaction thread
 # or a new MyApp::DB object can be created for independent transactions.
    
 my $newmodel = MyApp::Model->new( db => $db );      # autoinserts id
 $newmodel->{save}->{quick_dbfield1} = "new_value1";
 $newmodel->save;
 $newmodel->commit;   # or $newmodel->rollback;

 

DESCRIPTION

Perl Object-Oriented Framework attempts to provide an easier way to construct and deal with objects that map to entries in a database. To define a new class of object, in the example MyApp::Model, simply create tables in the database named "model" and "model_info". The names of these tables must be the lower case of your package name. model_info is optional and used for less-quick data associated w/ object.

Object::POOF and related classes require some conventions to be adhered to in order to reduce work and make object interfaces consistent.

An object package inheriting from Object::POOF need not define a constructor, instead, the generic Object::POOF constructor is inherited and used.

An object package inheriting from Object::POOF must declare a method 'init', which must first define the names of the tables associated with the object, and then calls SUPER::init, that is, Object::POOF::init(). (Right now this isn't used for anything, but it may be, so it should be called.)

The init() method MUST ALWAYS RETURN 1 UPON SUCCESS, otherwise the object will not be created and the inherited constructor will return undef.

An object package inheriting from Object::POOF also imports a wide array of data functions from Object::POOF::Data. These are designed to reduce the amount of work necessary to extract data from the database tables and to insert or replace data. Object::POOF::Data uses some nifty magic from Object::POOF::DB::Fields to determine whether these values should be quoted as strings or not, so you don't have to worry about it... just save whatever data you want and it should take care of it. If it is an incorrect data type (for example, a string saved to a numeric-type field), it will just ignore that field.

The object also gets the ability to call or create Shepherds (Object::POOF::Data::Shepherd), a machine class object that can find an instance of a given object based on field search criteria, or instantiate herds of objects from search criteria that can then be compared, crossed, ordered, etc.

BUGS

Rolling back newly created objects without saving

The only disadvantage to rollback after creating a new object is that the database auto_increment id counter has incremented. (Constructing a new object does an insert to get an ID to work with in the database thread.) With a big system creating a lot of objects temporarily that won't end up being saved, this could cause the auto_increment counter could run out of space over time. I'm not sure how to get around that, except making ID fields bigints.

MySQL is only supported database.

SEE ALSO

Object::POOF::Data(3), Object::POOF::Data::Shepherd(3), Object::POOF::DB(3), Object::POOF::App(3), Object::POOF::User(3)

AUTHOR

Copyright 2005 Mark Hedges <hedges@ucsd.edu>, CPAN: MARKLE

LICENSE

Released under the standard Perl license (GPL/Artistic).