# Model Inheritance, Mixins and Database design

**URL:** <https://www.ruby-forum.com/t/model-inheritance-mixins-and-database-design/54446>\
**Category:** Rails\
**Created:** [March 5, 2006, 11:49am UTC](https://www.ruby-forum.com/t/model-inheritance-mixins-and-database-design/54446 "2006-03-05T11:49:27Z")\
**Posts on this page:** 9\
**Page:** 1

<div class="post-metadata">

**Author:** ![acgreen](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/3bc359/32.png) [@acgreen](https://www.ruby-forum.com/u/acgreen)\
**Post date:** [March 5, 2006, 11:49am UTC](https://www.ruby-forum.com/t/model-inheritance-mixins-and-database-design/54446/1 "2006-03-05T11:49:27Z")

</div>

I’m trying to understand how best to use inheritance and wonder if  
Rail’s Single Table Inheritance model too restricting.

I have a model Companies who provide 1+ Services (HABTM relationship).

I assume the best way to represent this is with a series of child models  
representing each type of service : ServiceType \< Company.

I want to return lists of companys for each service and then link to  
show more details about them.

With the show method inherited from the Companies controller though  
changing the :id on the end of the url displays the next record in the  
companies table whereas I want to display the next company that also has  
that service available.

I was wondering a number of things:

1/ Should be overloading the find method in the ServiceType models.

2/ a/ Should I be trying to inherit the View for Company for ServiceType  
b/ by invoking Ruby’s super keyword somehow.

3/ Would it be better to have individual db table representing each  
ServiceType  
will the extra properiies that each ServiceType has in addition to the  
inherited Company properities require this

\_tony

---

<div class="post-metadata">

**Author:** ![acgreen](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/3bc359/32.png) [@acgreen](https://www.ruby-forum.com/u/acgreen)\
**Post date:** [March 5, 2006, 12:36pm UTC](https://www.ruby-forum.com/t/model-inheritance-mixins-and-database-design/54446/2 "2006-03-05T12:36:15Z")

</div>

Anthony

```
> I have a model Companies who provide 1+ Services (HABTM 

```

relationship).  
\> I assume the best way to represent this is with a series of child  
models  
\> representing each type of service : ServiceType \< Company.

I don’t understand your reasoning for using STI in this simple case.  
You’d be better served by a plain and standard habtm

```
 Company
     has_and_belongs_to_many :services

 Service
     has_and_belongs_to_many :companies

```

That would offer you simple access and manipulations :

```
 ikea = Company.find_by_name "ikea"
 is = ikea.services

 fs = Service.find_by_description "furnitures selling"
 companies_selling_furnitures = fs.companies

```

Alain

---

<div class="post-metadata">

**Author:** ![acgreen](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/3bc359/32.png) [@acgreen](https://www.ruby-forum.com/u/acgreen)\
**Post date:** [March 5, 2006, 1:21pm UTC](https://www.ruby-forum.com/t/model-inheritance-mixins-and-database-design/54446/3 "2006-03-05T13:21:46Z")

</div>

> ```
> Company
> has_and_belongs_to_many :services
> 
> Service
> has_and_belongs_to_many :companies
> 
> ```

This is what I started with.

> That would offer you simple access and manipulations :
> 
> ```
> ikea = Company.find_by_name "ikea"
> is = ikea.services
> 
> fs = Service.find_by_description "furnitures selling"
> companies_selling_furnitures = fs.companies
> 
> ```

How to I access  
company\_selling\_furniture.details

If I simply use an :id to have a url like

/companies/show/3

Then someone could just alter the url to

/company/show/4

which might not be a company selling furniture

But I also have access the additional properties of each service

```
company_selling_furniture.tables_for_sale

```

I’m trying to grasp how all the best practice concepts interlock

I’d assumed I should be generating models like

SellingFurnitureCompany \< Company

MakingFurnitureCompany \< Company

DeliveryCompany \< Company

Is this not the right approach ? Its just in the Pickaxe book they give  
the example of

JukeBoxSong \< Song  
KarokeSong \< Song

I assumed I should adopt a similar methodology.

\_tony

---

<div class="post-metadata">

**Author:** ![acgreen](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/3bc359/32.png) [@acgreen](https://www.ruby-forum.com/u/acgreen)\
**Post date:** [March 5, 2006, 2:03pm UTC](https://www.ruby-forum.com/t/model-inheritance-mixins-and-database-design/54446/4 "2006-03-05T14:03:54Z")

</div>

Anthony G. wrote:

> If I simply use an :id to have a url like
> 
> /companies/show/3
> 
> Then someone could just alter the url to
> 
> /company/show/4
> 
> which might not be a company selling furniture
> 
> But I also have access the additional properties of each service
> 
> ```
> company_selling_furniture.tables_for_sale
> 
> ```
> 
> I’m trying to grasp how all the best practice concepts interlock
> 
> I’d assumed I should be generating models like
> 
> SellingFurnitureCompany \< Company
> 
> MakingFurnitureCompany \< Company
> 
> DeliveryCompany \< Company
> 
> Is this not the right approach ? Its just in the Pickaxe book they give  
> the example of
> 
> JukeBoxSong \< Song  
> KarokeSong \< Song
> 
> I assumed I should adopt a similar methodology.
> 
> \_tony  
> This sounds like the right way to it, but what do you mean about the  
> id’s. Can you explain further?

joey\_\_  
[http://www.feedreed.com](http://www.feedreed.com)

---

<div class="post-metadata">

**Author:** ![acgreen](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/3bc359/32.png) [@acgreen](https://www.ruby-forum.com/u/acgreen)\
**Post date:** [March 5, 2006, 2:52pm UTC](https://www.ruby-forum.com/t/model-inheritance-mixins-and-database-design/54446/5 "2006-03-05T14:52:07Z")

</div>

On 3/5/06, Anthony G. [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote:

> so we have  
> they shouldn’t.  
> [http://lists.rubyonrails.org/mailman/listinfo/rails](http://lists.rubyonrails.org/mailman/listinfo/rails)

If you want to restrict users from viewing certain objects, you can  
build in an authentication system. Rails Recipes has a nice chapter  
on it. This would require you to do something like  
class User \< ActiveRecord::Base  
has\_one :company  
end

class Company \< ActiveRecord::Base  
belongs\_to :user  
end

Then in your list controller you do

def list  
@services = session[:user].company.services  
end

Since you’re accessing it all through the model, it’s protected (you  
still have to write the auth code of course, but that’s simple).

I highly recommend picking up Rails Recipes, it has a lot of stuff  
you’ll use.

Pat

---

<div class="post-metadata">

**Author:** ![acgreen](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/3bc359/32.png) [@acgreen](https://www.ruby-forum.com/u/acgreen)\
**Post date:** [March 5, 2006, 5:57pm UTC](https://www.ruby-forum.com/t/model-inheritance-mixins-and-database-design/54446/6 "2006-03-05T17:57:54Z")

</div>

> If you want to restrict users from viewing certain objects, you can  
> build in an authentication system.

You’ve misunderstodd the issue.  
I’m trying to understand why how child classes and inheritance interact  
in the rails framework.

\_tony

---

<div class="post-metadata">

**Author:** ![acgreen](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/3bc359/32.png) [@acgreen](https://www.ruby-forum.com/u/acgreen)\
**Post date:** [March 5, 2006, 2:23pm UTC](https://www.ruby-forum.com/t/model-inheritance-mixins-and-database-design/54446/7 "2006-03-05T14:23:15Z")

</div>

> This sounds like the right way to it, but what do you mean about the  
> id’s. Can you explain further?

Its the way Rails does its routing by default…  
:controller/:action/:id

In the list view its

/servicetype/list

in the show view its

/company/show/1

Now /servicetype/list will generate a list of links that point to the  
show controller passing also an :id value

the value :id that Rails generates points to the primary key of the  
record in the companies table

so we have

\<a href=/company/show/2"\>FirstCompanyWithThisServiceType  
\<a href=/company/show/5"\>SecondCompanyWithThisServiceType

clicking on the first link goes to a url  
/company/show/2  
that invokes the :controller =\> company, the :action =\> show and :id =\>  
2

if edit the url to /company/show/3 they’ll see the details of a company  
they shouldn’t.

\_tony

---

<div class="post-metadata">

**Author:** ![acgreen](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/3bc359/32.png) [@acgreen](https://www.ruby-forum.com/u/acgreen)\
**Post date:** [March 5, 2006, 7:36pm UTC](https://www.ruby-forum.com/t/model-inheritance-mixins-and-database-design/54446/8 "2006-03-05T19:36:49Z")

</div>

> I was responding to that. If I misinterpretted it, please explain it  
> better. The only way I can interpret that is “Users should not be  
> able to see the details of arbitrary companies.”

Keep in mind the thread is about inheritance and how it works with ORM.

The parent class maps to a table, an instance of that class mapping to a  
record in that table. The default list and show actions use :id to pick  
out a record in the table by its primary key.

Now a child class maps to the same table as their parent but the  
instance only maps to records based on particular criteria. It this  
situation using the combination of the show action and :id needs to be  
restricted from showing records not belonging to that child.

To try and keep the thread on track the things I asked originally asked  
were:

1/ Should I be overloading the find method in the ServiceType models.

2/ a/ Should I be trying to inherit the View for Company for ServiceType  
b/ by invoking Ruby’s super keyword somehow.

3/ Would it be better to have individual db table representing each  
ServiceType  
will the extra properiies that each ServiceType has in addition to the  
inherited Company properities require this

\_tony

---

<div class="post-metadata">

**Author:** ![acgreen](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/3bc359/32.png) [@acgreen](https://www.ruby-forum.com/u/acgreen)\
**Post date:** [March 5, 2006, 6:03pm UTC](https://www.ruby-forum.com/t/model-inheritance-mixins-and-database-design/54446/9 "2006-03-05T18:03:32Z")

</div>

On 3/5/06, Anthony G. [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote:

> > If you want to restrict users from viewing certain objects, you can  
> > build in an authentication system.
> 
> You’ve misunderstodd the issue.  
> I’m trying to understand why how child classes and inheritance interact  
> in the rails framework.

“if edit the url to /company/show/3 they’ll see the details of a  
company they shouldn’t.”

I was responding to that. If I misinterpretted it, please explain it  
better. The only way I can interpret that is “Users should not be  
able to see the details of arbitrary companies.”

Pat