[Catalyst] How to get default()-like behavior outside of the Root controller

[Trevor Leffler]

Hi Aristotle,

On 02/10/2015 02:39 AM, Aristotle Pagaltzis wrote:
> Hi Trevor,
>
> * Trevor Leffler <tleffler at uw.edu> [2015-02-07 00:00]:
>> Requests for non-extant paths fall through to Root->default(), which
>> gives the (undesired) default root view. I tried giving my top-level
>> controllers their own default() methods, but as others have found [1],
>> 'default() : Path' has precedence over 'index : Chained'.
>
>      package MyApp::Admin;
>      sub default : Chained('base') PathPart('') { ... }
>      package MyApp::Survey;
>      sub default : Chained('base') PathPart('') { ... }
>
> Note how you don’t specify any particular number of arguments. So it
> will match any number of trailing URL segments.
>
> I *think* this will work. I haven’t tried it.

I had tried this earlier with poor results.  However, I spun up a quick 
Foo catalyst project for testing and found that if I defined *all* the 
defaults()'s as :Chained in Root/Admin/Survey/etc. then they played well 
with my other methods, all of which are :Chained.  Once you go 
:Chained...  ;)

> However, contrary to what you claimed, splitting the app into multiple
> Cat apps joined together at the PSGI level *will* also work just fine.
> That will lead to requests for /admin/* always being dispatched to
> MyApp::Admin – which gets to have its own MyApp::Admin::Controller::Root
> with a `default` action that applies only to it. Likewise requests for
> /survey/* will always be dispatched to MyApp::Survey which equally has
> its own root controller with a `default` action that applies only to it.
> If the parts of your app are not closely related, this will be a more
> natural structure.

I don't think I spoke against splitting my app up, rather I just ignored 
that line of thought.  ;)  There may be too many 
relationships/dependencies across the major areas to split it up in the 
immediate future, but I can see some architectural advantages to doing 
so.  Simply the fact that I've got variations in template wrappers may 
be an indicator.

Thanks,
--Trevor

> Regards,
>