Re: might_have , has_many request

[prev] [thread] [next] [Date index for 2005/02/28]

--- Tony Bowden <tony-cdbitalk@xxxxx.xxx> wrote:

> On Mon, Feb 28, 2005 at 01:25:36PM -0800, Peter Speltz wrote:
> > I'd like it if might_have and has_many relationships took constraints. I've
> > patched my copies so i can use them the old way or like this:
> 
> I don't understand what your constraints actually do,

For might_have they essentially allow you to:
   * name the foreign key. Current version uses PK of class using might_have
     if i'm not mistaken..  _FK_ is special name in constraints for PK of 
     calling class (IE cd->primary_column in the standard example)
   * can might_have multi_keyed objects.  Just Specify all keys in constraints.
   * can might_have more than one of same class like in my example.

For has_many they just allow you to organize them and  saves a step when
accessing a limited group. Better example may be:
 Customer->has_many('dialup_accounts' => Account, { account_type => 'dialup'});
 Customer->has_many('dsl_accounts')   => Account, {account_type  => 'dsl' });

This really isn't that much of a big deal especially for display cause you can
limit when you access but for automating things it 
could be nice to have the extra meta info stored.

> but feel free to
> create your own Relationship classes and throw up on CPAN. Then people
> can just declare that they want has_many / might_have to use your versions
> rather than the built-ins if they want this behaviour.
> 

That's good idea. Any suggestions for names as they do just the same thing. 
can_have , has_many_constr ? 

> If you need anything in the default Relationships facotred out better to
> make it easier to subclass them, just shout.
> 

It would be nice to have a standard interface to pass args to all relationships
that will get remapped into the args slot of meta_info. This way people can add
custom meta info, like a flag or something without having to remap the args ,
which as far as i know means writing a new relationship. 

has_many expects args already and so does has_a. 
might_have doesn't. it slurps all args  into the import array. 



=====
pjs

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com