Thanks Simon for your revised feedback.
I still feel that making a managed-bean call in an iterator still
satisfies separation of view from model. I think its an issue of
how JSF builds components and how it uses a value derived in an
iterator, or I have not understood that well yet.
In any case, I managed to resolve my issue a bit differently. Here is
how, for those who may access this thread later on.
This was the original scenario, where the inner call to
#{officeBean.offices} would not take iterated value #{customer.id} from
#{customerBean.customers}
<ui:repeat var="customer" value="#{customerBean.customers}" >
Customer: #{customer.id} - #{customer.name} -
<ui:repeat var="office" value="#{officeBean.offices}">
Office: #{office.id}
</ui:repeat>
</ui:repeat>
So, on my 'customerBean' I added another method called
#{customerBean.officesByCustomers} which loads all offices for the
customers returned by #{customerBean.customers}, and returns a map, with
key as customer and values as a list of offices.
And, in the pseudo-code repeater I used it like this:
<ui:repeat var="customer" value="#{customerBean.customers}" >
Customer: #{customer.id} - #{customer.name} -
<ui:repeat var="office"
value="#{customerBean.officesByCustomers[customer.id]}">
Office: #{office.id}
</ui:repeat>
</ui:repeat>
Works as expected.
The drawback is that customerBean is pretty loaded up, but I guess that
guarantees the complete separation of view from service.
NOTE: this message is getting rejected by myfaces email server.. so I'm
not attaching any response form previous threads in this email ..
Thanks
MRather
