[Catalyst] Why scaffolding? Validation and Learning

[Matt S Trout]

Zbigniew Lukasiak wrote:
> Some more technical details.
> 
> The main idea of how the scaffolding should work is that we generate 
> only a skeleton of directories, nearly empty controllers and some config 
> stuff.  The generated controllers  only contain their package 
> declaration and a 'use base Catalyst::Example::Controller::InstantCRUD' 
> line. So there is not that much of generated code here.  After the 
> controllers are generated users are expected to copy the methods that 
> they want to modify from the Catalyst::Example::Controller::InstantCRUD 
> superclass - doing that they are taking responsibility for the copied 
> code.  This way the main Instant controller, which is just a standard 
> library and can be updated with standard means, is also an integral part 
> of the example.  The copying perhaps is not such a trivial step - but in 
> fact should be pretty natural for people working with Object Oriented 
> code - it's just overwriting of SUPER class methods. I am also planning 
> to make it more intuitive by the way of documentation.
> 
> This is the theory - in practice the config stuff required to have a 
> working application is a bit big, but we are working on it. 
> 
> A separate thing are the templates for the views, in the current CPAN 
> version they are treated in a similar way that the controllers - the 
> users are expected to just copy the generic templates from the 
> InstantCRUD directory and modify them to their liking.  There were 
> people on this list complaining about this and in the svn version now we 
> create the basic templates in the generated application.  This means 
> problems with the templates when upgrading the library, but perhaps we 
> need to treat the templates separately from the code and expect them to 
> be really quickly rewritten by the programmers.

We have an approach that allows standard templates to effectively be 
subclassed; once it goes to public announce I'd love to have your opinion on 
them since templates are the hardest issue for scaffolding and it's clear 
you've done a fair bit of thinking about it.

None of what I'm saying denigrates InstantCRUD - it's a very well-implemented 
Scaffold and I recommend it to people regularly. I just think it's possible to 
achieve the same things a better way.

-- 
      Matt S Trout       Offering custom development, consultancy and support
   Technical Director    contracts for Catalyst, DBIx::Class and BAST. Contact
Shadowcat Systems Ltd.  mst (at) shadowcatsystems.co.uk for more information

+ Help us build a better perl ORM: http://dbix-class.shadowcatsystems.co.uk/ +