Re: Implement Object Persistence (II)

[dragonchild]

I've submitted the following to jhuckaby (maintainer of DBM::Deep) as a bug report.

use DBM::Deep;
use Data::Dumper;
use Test::More no_plan => 1;

package Foo;

sub new { my $c = shift; bless { @_ }, $c }

package Bar;

sub new { my $c = shift; bless { @_ }, $c }

package main;

my $dbh = DBM::Deep->new( 'db.db' );

my $foo = Foo->new(
    bar => Bar->new( x => 1 ),
);

my $expected = Dumper( $foo );

$dbh->import( foo => $foo );

my $foo2 = $dbh->export( 'foo' );

my $actual = Dumper( $foo2 );

is( $actual, $expected );

[merlyn]

You mean, you want him to change the docs to make it very clear that DBM::Deep handles hashes and arrays only, not objects?

You're trying something that is documented not to work. Your bug, not DBM::Deep, unless you think it's a doc bug.

[dragonchild]

It's, at the very least a doc bug, in that the POD doesn't have this limitation in the CAVEATS. I originally thought it was an implementation bug, but the author responded and said it wasn't.

So sorry, but DBM::Deep doesn't handle objects (i.e. blessed hashes). It only supports anonymous hashes and arrays. Anything blessed will not be stored blessed. I considered adding this feature, but many classes don't appreciate having their hashes tied. Its on my to-do list for a future release, however.

I think it can be done, but that requires maintaining a separate metadata structure which may require a complete redesign of the file layout.

[merlyn]

Well, I read the very same docs you did (I presume), and I never got the impression that it would store objects. Interesting how we saw two entirely different things in the docs.

And frankly, I wouldn't want it to store objects. The design is clean... adding blessings alongside would clutter up the protocols needed to access the existing metadata. If you send it an "export" call, should it be DBM::Deep's export, or the native object export? That's the kind of confusion you'd get by permitting objects in the mix.