[Dbix-class] Multi-table join problems
Dan Horne
dan.horne at redbone.co.nz
Fri Jul 10 16:06:56 GMT 2009
On 11/07/2009, at 12:29 AM, wreis wrote:
> On Thu, Jul 9, 2009 at 8:19 PM, Dan Horne<dan.horne at redbone.co.nz>
> wrote:
>>
>> Hi
>> I have a structure like the one below - ep_config has many
>> ep_sections which
>> in turn has man ep_process_runs:
>> +----------------+ +-----------------+ +-----------------+
>>
>> | ep_config | | ep_section | | ep_process_run |
>> | | /| | /| |
>> | config_id (PK) |--------|-section_id (PK) |-----| process_id (PK) |
>> | config_name | \| config_id (FK) | \| section_id (FK) |
>> | | | | | |
>> +----------------+ +-----------------+ +-----------------+
>> I want to find all ep_process_runs where I know the config_name.
>> I've tried
>> the following:
>> my $rs = $schema->resultset('ETL::Pipeline::Schema::EpProcessRun')-
>> >search(
>> {
>> 'config_name' => $config_name
>> },
>> {
>> join => {'section_id' => 'config_id'},
>> order_by => 'config_name'
>> }
>> );
>
> You could do this as well:
>
> # if you have a unique constraint over config_name, you can use 'find'
> instead here too
> my $ep_config =
> $schema->resultset('ETL::Pipeline::Schema::EpConfig')->search_rs({
> config_name => $config_name
> }, { rows => '1' })->single;
> my $rs = $ep_config->ep_sections-
> >search_related_rs('ep_process_runs');
>
>> and I tried the condition
>> 'config_id.config_name' => $config_name
>> which gives
>> Can't locate object method "config_id"
>> The schema was reverse engineered sing DBIx::Class::Schema::Loader,
>> and it
>> names all belongs_to relationships using the column name specified
>> in the
>> FK/PK relationship. E.g. in ETL::Pipeline::Schema::EpSection it is
>> defined
>> as
>> __PACKAGE__->belongs_to(
>> "config_id",
>> "ETL::Pipeline::Schema::EpConfig",
>> { config_id => "config_id" },
>> );
>
> Change the relname "config_id" to something different like
> "ep_config", it's conflicting with 'config_id' column name.
That's what I did do in the end, but was trying to avoid this a)
because I don't want to change what DBix::Class::Schema::Loader
generates because then I'd need to change things every time I re-
generate the schema, and b) I was trying to operate under the
philosophy that I don't want to know about DBIC's implementation
details under the covers, but in the end, I needed tobe cognizant of
it. I don't know why DBIC::S::L names the relationships the same as
the columns - it actually threw me for a loop when I first started
using DBIC because none of the doc tutorials or examples did this, and
I found it very confusing.
thanks,
Dan
More information about the DBIx-Class
mailing list