[Catalyst] Feedback requested on HTML::FormHandler's new widget system

[Gerda Shank]

Matt Whipple wrote:
> From my perspective the output/rendering of the form data normally 
> represents a static view of the data, a single state which will 
> generally not require any return communication with the system which 
> renders it.  I therefore see a more independent treatment of widgets 
> the best option to allow those widgets to fit into the most places, 
> and be made the most accessible for future growth.  Rather than 
> treating the widgets as roles to be pulled into the form system, they 
> could be external objects with a common interface which are pushed 
> into and piped from. This aids in separation of concerns and could 
> lead to afar more flexible, robust widget framework which is more 
> conducive to collaboration and extension within and without the larger 
> system.
Interesting. The difficulty I see here is that the definition of various 
aspects of the presentation is currently done in the form/fields, and if 
I understand you correctly you are suggesting that that sort of 
definition should be done in some entirely different object. I am not 
really clear on how what you are proposing is different from a template 
system, which is basically an external system into which you push your 
variables, with a particular template defining the "interface".

You might be interested in the new 'result' object, which separates 
static from dynamic but in rather the opposite direction. Only the 
dynamic data is stored in the result objects. If you want to render from 
a result object it uses the static information in the form/fields to 
create the HTML. You can have multiple result objects which all refer to 
the same static form. This architecture is particularly suited to 
persistent forms.

Gerda