Class::AutoDB::Serialize - Serialization engine for Class::AutoDB -- MySQL only for now


This is a mixin class that enables objects to be serialized and stored in a database as independent entities, and later fetched one-by-one or in groups. Whether fetched individually or in groups, the original shared object structure is preserved. It's not necessary for the object to also from Class::AutoClass, although the examples here all do.

Define class that inherits from Class::AutoDB::Serialize

Person class with attributes 'name', 'sex', 'hobbies', and 'friends', where 'friends' is a list of Persons.

 package Person;
 use Class::AutoClass;
 use Class::AutoDB::Serialize;
 @ISA=qw(Class::AutoClass Class::AutoDB::Serialize); 
 @AUTO_ATTRIBUTES=qw(name sex hobbies friends);

Create and store some objects

 use DBI;
 use Class::AutoDB::Serialize;
 use Person;
 my $dbh=DBI->connect('dbi:mysql:database=ngoodman;host=localhost');
 my $joe=new Person(-name=>'Joe',-sex=>'male',
                    -hobbies=>['mountain climbing', 'sailing']);
 my $mary=new Person(-name=>'Mary',-sex=>'female',
                     -hobbies=>['hang gliding']);
 my $bill=new Person(-name=>'Bill',-sex=>'male',
                     -hobbies=>['cooking', 'eating', 'sleeping']);
 # Set up friends lists
 # Store the objects
 # Print their object id's so you can fetch them later
 for my $obj ($joe, $mary, $bill) {
   print 'name=', $obj->name, ' oid=', $obj->oid, "\n";

Fetch the objects

Assume that the oid's are passed as command line arguments

 my @oids=@ARGV;
 my $joe=Class::AutoDB::Serialize::fetch($ARGV[0]);
 my $mary=Class::AutoDB::Serialize::fetch($ARGV[1]);
 my $bill=Class::AutoDB::Serialize::fetch($ARGV[2]);
 # Print the objects' attributes
 for my $obj ($joe, $mary, $bill) {
   print 'oid=', $obj->oid, "\n";
   print 'name=', $obj->name, "\n";
   print 'sex=', $obj->sex, "\n";
   print 'hobbies=', join(', ',@{$obj->hobbies}), "\n";
   print 'friends=',"\n";
 for my $friend (@{$obj->friends}) {
   print ' oid=', $friend->oid, ', ';
   print 'name=', $friend->name, "\n";
 print "----------\n";
 # Change an attribute in each object to demonstrate 
 # that shared structure is preserved
 for my $obj ($joe, $mary, $bill) {
   $obj->name($obj->name.' Changed');
 for my $obj ($joe, $mary, $bill) {
   print 'oid=', $obj->oid, "\n";
   print 'changed name=', $obj->name, "\n";
   print 'friends=',"\n";
   for my $friend (@{$obj->friends}) {
     print ' oid=', $friend->oid, ', ';
     print 'changed name=', $friend->name, "\n";
   print "----------\n";


This is a mixin class that implements the serialization and data storage engine for Class::AutoDB. Objects that inherit from this class can be serialized and stored in a database as independent entities. This only works for objects implemented as HASHes in the usual Perl way.

What distinguishes Class::AutoDB::Serialize from the many other excellent Perl serialization packages (eg, Data::Dumper, Storable, YAML) is that we serialize objects as independent entities existing within a large network, rather than serializing entire networks as a whole. When an object is being serialized, other “auto-serializable” objects that are encountered are not serialized then and there; instead a placeholder object called an Oid (short for object identifier) is emitted into the serialization stream. When the object is later fetched, the placeholders are not immediately fetched; instead each placeholder is fetched transparently when the program invokes a method on it (this is accomplished via an AUTOLOAD mechanism).

The purpose of all this is to make it easy for Perl programs to operate on large databases of objects. Objects can be created, stored, and later fetched. If the object points to other objects, they will be fetched when needed. New objects can be created, connected to the network of existing objects, and stored.