[Dbix-class] load_classes and common base class

[Ash Berlin]

Bill Moseley wrote:
> On Sat, Apr 21, 2007 at 03:19:14PM +0100, Ash Berlin wrote:
>> The way I do things is;
> 
> Ok, so your result sources just inherit from a base class.
> 
> 
> Ok, so this has me wondering:
> 
>>     # $self->resultset_class('My::ResultSet');
> 
> I noticed that in the cookbook.  Are you using a single class to
> define commonly used result sets?  Seems like one would want custom
> result sets to be specific to a given table class.

In the case where i use the above, I actually just use that as a base 
class for a group of related sources - in my case tables that store test 
results. Since each test needs slightly different data to be stored, as 
well as common data, i do it that way.

> 
> Where is resultset_class documented?

find . -name *.pm -or -name *.pod | xargs grep -H resultset_class | less
...
./lib/DBIx/Class/ResultSource.pm:=head2 resultset_class

There

> 
> I think an example would help clear things up for me quite a bit:
> 
> 
> In CDBI I create extra methods in the table classes as needed to
> extend those classes.  Sometimes these are instance methods and other
> times class methods.
> 
> 
> For example, I have the common setup where when someone signs up for
> an account they get sent a confirmation email that they must use to
> complete the registration process.
> 
> I have a table account_confirmation that tracks these pending accounts.
> I also have a users table (called "person") where I store their signup
> data.
> 
> People will sign up for an account, but never complete the process by
> responding to the account confirmation email, so I have to
> periodically delete those pending accounts.
> 
> In CDBI I have a class methods that does that:
> 
>     App::AccountConfirmation->clean;
> 
> And a feature (or problem) with CDBI is that the $dbh can be found
> from the class using Foo->db_Main.  Don't need to pass $schema in, for
> example.
> 
> That means I can do:
> 
> package AccountConfirmation;
> 
> sub clean {
>     my $self = shift;
> 
>     # Deleting a person will cascade delete account_confirmation
> 
>     my $sql = <<'';
>         delete from person where id in
>         (
>             SELECT
>                 p.id
>             FROM
>                 person p
>                 JOIN account_confirmation a ON (p.id = a.person)
>             WHERE
>                 a.created_time < now() - interval '1 day'
>         )
> 
>     return $self->db_Main->do( $sql );
> }
> 
> So, how do I implement something like that in DBIC?
> 
> 
> Is this where ResultSetManager comes in?
> 
> 

Could be done with RSM yes, or use resultset_class and make it a method 
on that class. Then:

# if using RSM
# sub clean: ResultSet {
sub clean {
   my ($rs) = @_;

   # This might not be exactly right mind...
   $rs->search(
     {'account_confirmation.created_time' => \'< now() - interval '1 
day''},
     { join => 'account_confirmation'}
   )->delete();
}

Make more sense?