Rails core - Aug 2006 - Super DRY models - Auto-associations and validations

I''ve written an ActiveRecord extension that allows Associations to be
automatically generated at runtime without requiring "has_many" and
"belongs_to" to be defined in the class. The use of has_many vs
has_one is
determined by the pluralisation of the method call.

Additionally, it uses the :null => false attribute of columns to provide
validates_presence_of automatically to models too.

If any of this code is relevant for Core Rails, let me know and I''ll
extract
it from the gem and create a patch for core. The gem has received a good
reception on the Ruby + RoR forums.

http://drnicwilliams.com/2006/08/07/ann-dr-nics-magic-models/

Cheers
Nic

-- 
Dr Nic Williams
http://www.drnicwilliams.com - Ruby/Rails blog
skype: nicwilliams
(m) +31 62 494 8552
(p) +61 7 3305 7522 (finds me anywhere in the world)
(f) +61 7 3305 7572 (sends fax to my email)


_______________________________________________
Rails-core mailing list
Rails-core@lists.rubyonrails.org
http://lists.rubyonrails.org/mailman/listinfo/rails-core

Hi Nic,

I for one think this is incredibly cool, and would love to see it in  
Rails 1.2 (as part of core)...

Thanks!
-- Ernie P.
On Aug 8, 2006, at 3:33 AM, Nic Williams wrote:
> I''ve written an ActiveRecord extension that allows Associations to
> be automatically generated at runtime without requiring
"has_many"
> and "belongs_to" to be defined in the class. The use of has_many
vs
> has_one is determined by the pluralisation of the method call.
>
> Additionally, it uses the :null => false attribute of columns to  
> provide validates_presence_of automatically to models too.
>
> If any of this code is relevant for Core Rails, let me know and  
> I''ll extract it from the gem and create a patch for core. The gem
> has received a good reception on the Ruby + RoR forums.
>
> http://drnicwilliams.com/2006/08/07/ann-dr-nics-magic-models/
>
> Cheers
> Nic
>
> -- 
> Dr Nic Williams
> http://www.drnicwilliams.com - Ruby/Rails blog
> skype: nicwilliams
> (m) +31 62 494 8552
> (p) +61 7 3305 7522 (finds me anywhere in the world)
> (f) +61 7 3305 7572 (sends fax to my email)
>
> _______________________________________________
> Rails-core mailing list
> Rails-core@lists.rubyonrails.org
> http://lists.rubyonrails.org/mailman/listinfo/rails-core


_______________________________________________
Rails-core mailing list
Rails-core@lists.rubyonrails.org
http://lists.rubyonrails.org/mailman/listinfo/rails-core

I would agree that this is a nice way to get started with your models,
but once you need to do anything beyond the basics, like association
extensions for example, even simply renaming the association, then
you''ll have to define the association anyway.

Nice idea, but I''d rather see this remain as a plugin instead of users
getting used to the fact that you don''t have to define model
associations or provide validations.

If all this did was generate the base model with the associations and
validations then that would definitely relieve some initial
development time and possibily worthy of core inclusion.

Andrew

On 8/14/06, Dr. Ernie Prabhakar <drernie@opendarwin.org>
wrote:
>
> Hi Nic,
>
> I for one think this is incredibly cool, and would love to see it in Rails
> 1.2 (as part of core)...
>
> Thanks!
> -- Ernie P.
>
> On Aug 8, 2006, at 3:33 AM, Nic Williams wrote:
> I''ve written an ActiveRecord extension that allows Associations to
be
> automatically generated at runtime without requiring "has_many"
and
> "belongs_to" to be defined in the class. The use of has_many vs
has_one is
> determined by the pluralisation of the method call.
>
> Additionally, it uses the :null => false attribute of columns to provide
> validates_presence_of automatically to models too.
>
> If any of this code is relevant for Core Rails, let me know and
I''ll extract
> it from the gem and create a patch for core. The gem has received a good
> reception on the Ruby + RoR forums.
>
> http://drnicwilliams.com/2006/08/07/ann-dr-nics-magic-models/
>
> Cheers
> Nic
>
> --
> Dr Nic Williams
>  http://www.drnicwilliams.com - Ruby/Rails blog
> skype: nicwilliams
> (m) +31 62 494 8552
> (p) +61 7 3305 7522 (finds me anywhere in the world)
> (f) +61 7 3305 7572 (sends fax to my email)
> _______________________________________________
> Rails-core mailing list
> Rails-core@lists.rubyonrails.org
> http://lists.rubyonrails.org/mailman/listinfo/rails-core
>
> _______________________________________________
> Rails-core mailing list
> Rails-core@lists.rubyonrails.org
> http://lists.rubyonrails.org/mailman/listinfo/rails-core
>
>
>

I like this, and i believe most people would.
I can''t say for the core team, but i think you sould create a patch
and post it on trac. The automatically validation can be improved to
more than just validates_presence_of as well, others people might help
with that and then it would be easier to apply the patch on the trunk
;)

Nice job, Nic.

Cheers!

Rodrigo Kochenburger

On 8/14/06, Dr. Ernie Prabhakar <drernie@opendarwin.org>
wrote:
>
> Hi Nic,
>
> I for one think this is incredibly cool, and would love to see it in Rails
> 1.2 (as part of core)...
>
> Thanks!
> -- Ernie P.
>
> On Aug 8, 2006, at 3:33 AM, Nic Williams wrote:
>
> I''ve written an ActiveRecord extension that allows Associations to
be
> automatically generated at runtime without requiring "has_many"
and
> "belongs_to" to be defined in the class. The use of has_many vs
has_one is
> determined by the pluralisation of the method call.
>
> Additionally, it uses the :null => false attribute of columns to provide
> validates_presence_of automatically to models too.
>
> If any of this code is relevant for Core Rails, let me know and
I''ll extract
> it from the gem and create a patch for core. The gem has received a good
> reception on the Ruby + RoR forums.
>
> http://drnicwilliams.com/2006/08/07/ann-dr-nics-magic-models/
>
> Cheers
> Nic
>
> --
> Dr Nic Williams
>  http://www.drnicwilliams.com - Ruby/Rails blog
> skype: nicwilliams
> (m) +31 62 494 8552
> (p) +61 7 3305 7522 (finds me anywhere in the world)
> (f) +61 7 3305 7572 (sends fax to my email)
>
> _______________________________________________
> Rails-core mailing list
> Rails-core@lists.rubyonrails.org
> http://lists.rubyonrails.org/mailman/listinfo/rails-core
>
> _______________________________________________
> Rails-core mailing list
> Rails-core@lists.rubyonrails.org
> http://lists.rubyonrails.org/mailman/listinfo/rails-core
>
>
>

-- 
Rodrigo Kochenburger
<divoxx at gmail dot com>
Linkedin professional profile: http://www.linkedin.com/in/rodrigok

> I''ve written an ActiveRecord extension that allows Associations to
be
> automatically generated at runtime without requiring "has_many"
and
> "belongs_to" to be defined in the class. The use of has_many vs
has_one is
> determined by the pluralisation of the method call.
Very cool stuff, Nic. I think it would be a great app/plugin to use
against databases that follow Rails conventions but that does not
necessarily have a Rails application already.

I don''t think its a good fit for core, though. Primarily because it
splinters the overview. I don''t have any models in any of my real
applications where the automatically defined rules would be
sufficient. And when they''re not, it means that my validates_format_of
will have to be defined explicitly while the validates_presence_of is
magic. Or that my has_many :comments, :order => "created_at DESC"
has
to be explicit, but my has_one :author is automatic.

But again, please do keep pushing on with this. Just because its not a
good fit for core, doesn''t mean it ain''t cool. Rock on.
-- 
David Heinemeier Hansson
http://www.loudthinking.com -- Broadcasting Brain
http://www.basecamphq.com   -- Online project management
http://www.backpackit.com   -- Personal information manager
http://www.rubyonrails.com  -- Web-application framework

> If all this did was generate the base model with the associations and
> validations then that would definitely relieve some initial
> development time and possibily worthy of core inclusion.
+1 for addition to script/generate model/scaffold/etc.

would be to use via a more standalone script like script/console (e.g.
script/sql ) where I can connect to any db and immediately get models and
benefits of AR... then perhaps I really won''t have to enter the mysql
prompt
anymore.



On 8/15/06, Kyle Maxwell <kyle@kylemaxwell.com>
wrote:
>
> > If all this did was generate the base model with the associations and
> > validations then that would definitely relieve some initial
> > development time and possibily worthy of core inclusion.
>
> +1 for addition to script/generate model/scaffold/etc.
> _______________________________________________
> Rails-core mailing list
> Rails-core@lists.rubyonrails.org
> http://lists.rubyonrails.org/mailman/listinfo/rails-core
>

_______________________________________________
Rails-core mailing list
Rails-core@lists.rubyonrails.org
http://lists.rubyonrails.org/mailman/listinfo/rails-core

On 8/14/06, Andrew Kaspick <akaspick@gmail.com>
wrote:
>
>
> If all this did was generate the base model with the associations and
> validations then that would definitely relieve some initial
> development time and possibily worthy of core inclusion.
>
It can''t really be included with script/generate model since it
bases what it would generate on the migration that would not be created yet.
Some sort of script you could run after the model was created and the
migrations were done that would add the code would be pretty nice tho. Maybe
even something that you could run multiple times that would (and this might
be hard) not step on your toes if you had already modified the model.

Martin


_______________________________________________
Rails-core mailing list
Rails-core@lists.rubyonrails.org
http://lists.rubyonrails.org/mailman/listinfo/rails-core

@Andrew and Kyle: I''ve started a magic_model_generator that will be
released
after a foreign key support patch. That''ll be a wonderful tool. Not
very
magical, but useful... non-magically.

@DHH and Andrew : I agree with that - in a normal app development,
you''ll
always want a model class and explicit associations - eventually.

Its probably a style preference for agile developers as to whether they are
happy to immediately start work with derived associations (like the derived
attr_accessors for fields in ARs now) instead of explicitly defining them.
"Opinions" are great; so is choice. :)

Over time, we''ll start to get feedback on people''s development
philosophies/strategies, for new Rails apps, Rails apps on legacy DBs, and
non-Rails apps, etc.

@Rodrigo - Apparently post-Railsconf (read, post-Dave Thomas talk), some
work was done on auto-validations by some people which is more magical that
the simple validations that are available now; so I''ll include those
too;
e.g. validates_uniqueness_of if its a sole-index, etc. Wonderful stuff!

Cheers
Nic

-- 
Dr Nic Williams
http://www.drnicwilliams.com - Ruby/Rails blog
skype: nicwilliams
(m) +31 62 494 8552
(p) +61 7 3102 3237 (finds me anywhere in the world)
(f) +61 7 3305 7572 (sends fax to my email)


_______________________________________________
Rails-core mailing list
Rails-core@lists.rubyonrails.org
http://lists.rubyonrails.org/mailman/listinfo/rails-core

> Its probably a style preference for agile developers as to whether they are
> happy to immediately start work with derived associations (like the derived
> attr_accessors for fields in ARs now) instead of explicitly defining them.
> "Opinions" are great; so is choice. :)
I agree. With Rails, the split is that opinions go into core and
options go into plugins.
-- 
David Heinemeier Hansson
http://www.loudthinking.com -- Broadcasting Brain
http://www.basecamphq.com   -- Online project management
http://www.backpackit.com   -- Personal information manager
http://www.rubyonrails.com  -- Web-application framework