[Catalyst] Re: [RHTMLO] RFC: Catalyst::Controller::RHTMLO

[Jason Gottshall]

Jason Gottshall wrote:
> John Siracusa wrote:
>> On Mon, Jan 26, 2009 at 1:43 PM, Jason Gottshall 
>> <jgottshall at capwiz.com> wrote:
>>> Caching sounds wise. The only kink I see is that my controller allows 
>>> for
>>> multiple instances of the same form.
>>
>> The "same form" meaning what?  If you store forms keyed by an
>> arbitrary name string, you can have as many forms as you want so long
>> as the names are unique.
> 
> My implementation is designed to instantiate subclasses of 
> Rose::HTML::Form based on the class name, which would likely also be the 
> cache key, hence the difficulty in having multiple instances of the same 
> form.
> 
> Honestly, the only use case I can actually envision for having the same 
> form on a page more than once is maybe displaying a simple search form 
> both above and below a set of results, and even then you wouldn't 
> actually need two instances of the class, you'd just need to render the 
> same object twice. Unless anyone can see a use case I'm missing, I think 
> the point is moot, and I will just remove that whole "same form twice" 
> concept from the module and be done with it.

I've been thinking some more about the caching issue, and I'm concerned 
about the fact that rose form objects are mutable beyond just setting 
values for the fields. Any code that works with a form object can 
add/delete fields and subforms, monkey with validation routines, etc. 
Unless I can reset a form object to it's "new()-like" state before I 
reuse it, it seems like I really need a whole new instance to work with 
on each request. I haven't benchmarked yet; is there really that much 
overhead in constructing these form objects?

-- 
Jason Gottshall
jgottshall at capwiz.com