[Dbix-class] Nested create with relationships

[Matt S Trout]

On Thu, Apr 17, 2008 at 08:13:40PM -0700, Kate Yoak wrote:
> Hi everyone!
> 
> I am new here - a cdbi convert.
> 
> I am creating a record with its has_many relationship and, though  
> create succeeds, I wind up with data in the has_many table without  
> the foreign key.
> 
> Here is the data that goes into create:
> 
> $VAR1 = {
>           'features' => [
>                           {
>                             'feature_id' => undef,
>                             'feature' => 'kdkdk',
>                             'ord' => '1'
>                           },
>                           {
>                             'feature_id' => undef,
>                             'feature' => 'kdkdkdk',
>                             'ord' => '2'
>                           },
>                           {
>                             'feature_id' => undef,
>                             'feature' => 'jjjj',
>                             'ord' => '3'
>                           }
>                         ],
>           'sku' => 'cccc',
>           'name' => 'kdkdk',
>           'description' => 'kdkdkdk'
>         };
> 
> The relationship's foreign key is product_id - it is created as a  
> result of the main insert - but is not propagated into the feature  
> table.  Here are the inserts:

That's ... odd. Can you send us the classes? In the case of the systems
involved it should do the insert of the main record first, then the
has_many (which, see below, it -does-), but somehow it doesn't have the PK
afterwards.

> In addition, why are there so many queries? What is count(*) being  
> selected for and why are we doing one insert at a time instead of a  
> bulk insert? Is that something I should improve on?

The extra SELECTs are because DBIC hasn't worked out that the has_many
records you've provided are definitely new and is going to check that
they don't already exist so it doesn't accidentally cause a unique
key violation.

In this case it should probably spot that there's no possible unique key
to violate and not try the SELECTs; given that we could -probably- also
do a bulk insert, but most databases don't let you get an auto-inc key
in bulk insert so I'm not sure how we'd distinguish whether that's safe
or not?

-- 
      Matt S Trout       Need help with your Catalyst or DBIx::Class project?
   Technical Director                    http://www.shadowcat.co.uk/catalyst/
 Shadowcat Systems Ltd.  Want a managed development or deployment platform?
http://chainsawblues.vox.com/            http://www.shadowcat.co.uk/servers/