https://issues.apache.org/jira/browse/WICKET-2298

-igor

On Mon, Jun 1, 2009 at 11:45 AM, Igor Vaynberg <[email protected]> wrote:
> yeah, this is a problem with the design of the resource resolution.
> unfortunately we cannot fix until 1.5. here is an email thread where
> some possible future solutions were discussed.
>
> http://markmail.org/thread/lj3luznjnvbun72r
>
> i didnt notice a jira issue for this so i will add one.
>
> -igor
>
> On Mon, Jun 1, 2009 at 11:08 AM, Carlos Pita <[email protected]> wrote:
>> Hi all,
>>
>> we have a site where some pages/components implement variant V. Now
>> there is a requirement to sell a mainly reskinned version S of this
>> site to a customer. So my first thought was to implement this
>> site-wide "variant" S as a style. But soon I realised that _V wouldn't
>> match variant V for site S. That is, that there is no default site for
>> variations of a component. So, if I had:
>>
>> MyComponent
>> MyComponent_V
>>
>> I'm now forced to duplicate MyComponent_V in MyComponent_S_V even if
>> there is no difference at all for that component in both sites (which
>> is the common case).
>>
>> Also, as ComponentStringResourceLoader directly instantiates
>> ResourceNameIterator, there is no simple hook to customise the
>> resolution sequence. One should reimplement
>> ComponentStringResourceLoader and its subclass
>> ClassStringResourceLoader in order to get such simple change in the
>> resolution algorithm. This is done mostly by copy&paste.
>>
>> I think that:
>>
>>  i. style and variations are not treated very orthogonally,
>> MyComponent above matches both the default style and S, but
>> MyComponent_V matches only the default site. IMO this violates the
>> rule of least surprise, I would expected: MyComponent_V_S then
>> MyComponent_V then MyComponent_S then MyComponent. Anyway, I agree
>> this is arguable, because of the ambiguity between MyComponent_V and
>> MyComponent_S: it's not very clear which should go first. But in this
>> case, the impossibility to define a sequence that is good for everyone
>> reinforces next point (ii).
>>
>>  ii. ResourceNameIterator should be configurable (that is, I expect
>> some ioc here). Also style and variation shouldn't be merged at one
>> single parameter at this point; this falls under ResourceNameIterator
>> responsability instead.
>>
>> What do you think?
>>
>> Kind regards
>> -Carlos
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: [email protected]
>> For additional commands, e-mail: [email protected]
>>
>>
>

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to