Hi
The point is there is an implementation or ui:repeat and UIData that
behaves as expected by the spec without any additional changes. The only
difference we have between both implementations is that one has a bug in
that part and the other does not. The fix is just implement
setDataModel(...), getDataModel(...) and other related methods in a way
that assume the same component instances is used many times in the same
phase by different "virtual" components.
I do not see a use case where the fix could break older code, or at least
there is no report suggesting that the current ui:repeat implementation
contains a compatibility bug.
But if we talk about c:forEach tag, that's another story, because that
component has been proven to be really hard to fix, and any attempt to fix
it breaks backwards compatibility. I do not how, but the users always found
new creative ways to break that component, but I think the latest version
we have in MyFaces Core 2.2.9 is the best fix so far.
The fix is just use a map to store the DataModel instances, and use the
container clientId to get the right instance. That sounds better than add a
Repeater interface.
regards,
Leonardo Uribe
2016-01-15 9:54 GMT-05:00 arjan tijms <arjan.tijms_at_gmail.com>:
> It's a minimal thing to "mark" a component as being an iterating/repeating
> component. One use for that is that if needed child components can take a
> special action if they detect they're nested inside such component.
>
> Tightening up the spec specifically for ui:repeat and h:dataTable is more
> difficult. The problem I see (and please correct me if I'm wrong) is that
> these are older components. These components may contain certain bugs and
> those bugs in some cases may even be a spec violation, but these bugs have
> been there for a long time.
>
> At some point the bug, essentially, becomes a kind of de facto spec, since
> there's older code depending on it. Fixing the bug means breaking the older
> code, which means breaking the rule that existing code can never be broken.
>
> Now I'm of course not saying that no bug should ever be fixed, but that
> fixing (obvious) bugs is not always easy for a project with such strict
> compatibility rules.
>
> Nevertheless, if someone knows how we can tighten up the spec without
> breaking existing code, it may be worthwhile to look into that.
>
>
>
>
>
> On Fri, Jan 15, 2016 at 3:28 PM, Kito Mann <kito.mann_at_virtua.com> wrote:
>
>> Just to be clear, we're proposing an interface so that people can write
>> custom ui:repeat components, but not tightening up the spec so that both
>> implementations behave the same?
>>
>> ___
>>
>> Kito D. Mann | @kito99 | Author, JSF in Action
>> Web Components, Polymer, JSF, PrimeFaces, Java EE, and Liferay training
>> and consulting
>> Virtua, Inc. | virtua.tech
>> JSFCentral.com | @jsfcentral
>> +1 203-998-0403
>>
>> * Listen to the Enterprise Java Newscast: *http://
>> <http://blogs.jsfcentral.com/JSFNewscast/>enterprisejavanews.com
>> <http://ww.enterprisejavanews.com>*
>>
>>
>> On Thu, Jan 14, 2016 at 10:29 AM, manfred riem <manfred.riem_at_oracle.com>
>> wrote:
>>
>>> Hi all,
>>>
>>> Any proposal for changes needs to take into account the behavior that
>>> ui:repeat has had for
>>> a long time, whether that is explicitly mentioned in the specification
>>> or not. As otherwise
>>> you would break existing applications.
>>>
>>> This proposal looks like it does not break anything, but obviously it
>>> actually needs to be
>>> tested to see if it works. So please continue and deliver a fix/patch
>>> that is/can be tested.
>>>
>>> Thanks!
>>>
>>> Kind regards,
>>> Manfred Riem
>>>
>>>
>>> On 1/13/16, 3:10 PM, arjan tijms wrote:
>>>
>>> Hi,
>>>
>>> On Wed, Jan 13, 2016 at 5:53 PM, Cagatay Civici <
>>> cagatay.civici_at_primetek.com.tr> wrote:
>>>
>>>> Following is from Mojarra that is used by UIData and UIRepeat;
>>>>
>>>> private Boolean isNestedWithinIterator() {
>>>> if (isNested == null) {
>>>> UIComponent parent = this;
>>>> while (null != (parent = parent.getParent())) {
>>>> if (parent instanceof UIData ||
>>>> parent.getClass().getName().endsWith("UIRepeat")) {
>>>> isNested = Boolean.TRUE;
>>>> break;
>>>> }
>>>> }
>>>> ...
>>>>
>>>> [...]
>>>> In terms of Spec, an interface like Repeater should solve it where we
>>>> at PF can implement as well for our components.
>>>>
>>>
>>> We have had comparable issues with that same fragment. I like the
>>> Repeater proposal. If we just let it be a marker interface and then have
>>> both UIData and UIRepeat implement it, then the check can become:
>>>
>>> if (parent instanceof Repeater ||
>>> parent.getClass().getName().endsWith("UIRepeat")) {
>>> isNested = Boolean.TRUE;
>>> break;
>>> }
>>>
>>> It still has the class name check, but I think that's a small price for
>>> the sake of backwards compatibility.
>>>
>>> This would be a relatively small change, in terms of coding effort
>>> trivial almost. Cagatay, would the above change fully fix the issue for you?
>>>
>>> Parallel to this, does anyone think it's a good idea to investigate if
>>> it's necessary/useful to introduce a new iterating (base) component? I took
>>> a few looks at the existing code of UIData for some of the above issues and
>>> I fear that fixing those while at the same time being 100% backwards
>>> compatible is undoable.
>>>
>>> Kind regards,
>>> Arjan Tijms
>>>
>>>
>>>
>>>
>>>
>>>> On Wednesday 13 January 2016 at 16:27, arjan tijms wrote:
>>>>
>>>> Hi,
>>>>
>>>> Interesting, indeed. There's quite an amount of issues in the spec
>>>> tracker that concern ui:repeat and the related h:dataTable. I planned to do
>>>> a JSF 2.3 wishlist part II, but never came around to doing that.
>>>>
>>>> See:
>>>>
>>>> - https://java.net/jira/browse/JAVASERVERFACES_SPEC_PUBLIC-236
>>>> - UIData needs review on a couple of fronts
>>>> https://java.net/jira/browse/JAVASERVERFACES_SPEC_PUBLIC-935
>>>> - Alignment/extending of iterating ui components
>>>> https://java.net/jira/browse/JAVASERVERFACES_SPEC_PUBLIC-963
>>>> - Provide a component for iterating over hierarchical data
>>>> https://java.net/jira/browse/JAVASERVERFACES_SPEC_PUBLIC-965
>>>> - ui:repeat requires document "begin" and "end" properties
>>>> https://java.net/jira/browse/JAVASERVERFACES_SPEC_PUBLIC-892
>>>> - add "maxRepeats" attribute to ui:repeat.
>>>> https://java.net/jira/browse/JAVASERVERFACES_SPEC_PUBLIC-1249
>>>> - Add a styleClass attribute to h:column
>>>> https://java.net/jira/browse/JAVASERVERFACES_SPEC_PUBLIC-217
>>>> - Inconsistent column styles handling with h:dataTable
>>>> https://java.net/jira/browse/JAVASERVERFACES_SPEC_PUBLIC-627
>>>> - ui:repeat varStatus incompletely specified
>>>> https://java.net/jira/browse/JAVASERVERFACES_SPEC_PUBLIC-609
>>>> - Add rowStatePreserved property to UIRepeat, exactly the same as
>>>> UIData https://java.net/jira/browse/JAVASERVERFACES_SPEC_PUBLIC-1263
>>>> - UIData (and related classes) doesn't respect contract for
>>>> UIComponent process
>>>> https://java.net/jira/browse/JAVASERVERFACES_SPEC_PUBLIC-1120
>>>> - UIRepeat supports Collection
>>>> https://java.net/jira/browse/JAVASERVERFACES_SPEC_PUBLIC-1103
>>>> - Add support for condition check like regular for-loop in
>>>> ui:repeat
>>>> https://java.net/jira/browse/JAVASERVERFACES_SPEC_PUBLIC-1102
>>>> - Have DataModel implementations registerable
>>>> https://java.net/jira/browse/JAVASERVERFACES_SPEC_PUBLIC-1078
>>>> - h:datatable component to handle multiple entities lists
>>>> https://java.net/jira/browse/JAVASERVERFACES_SPEC_PUBLIC-713
>>>> - UIRepeat et al are not specified
>>>> https://java.net/jira/browse/JAVASERVERFACES_SPEC_PUBLIC-704
>>>> - Ajax inside a DataTable (getClientId append rowIndex)
>>>> https://java.net/jira/browse/JAVASERVERFACES_SPEC_PUBLIC-865
>>>>
>>>> Perhaps some of the inconsistencies Cagatay found are related to any of
>>>> those?
>>>>
>>>> Kind regards,
>>>> Arjan Tijms
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> On Wed, Jan 13, 2016 at 3:15 PM, Kito Mann <kito.mann_at_virtua.com>
>>>> wrote:
>>>>
>>>> Hello Cagatay,
>>>>
>>>> I noticed you just added a PrimeFaces repeat component
>>>> <https://github.com/primefaces/primefaces/issues/999> citing "inconsistencies
>>>> between ui:repeat implementations of mojarra and myfaces with our
>>>> components". Is there something we need to do with the spec to ensure that
>>>> the two implementations of ui:repeat behave more consistently?
>>>> ___
>>>>
>>>> Kito D. Mann | @kito99 | Author, JSF in Action
>>>> Web Components, Polymer, JSF, PrimeFaces, Java EE, and Liferay training
>>>> and consulting
>>>> Virtua, Inc. | virtua.tech
>>>> JSFCentral.com | @jsfcentral
>>>> +1 203-998-0403 <%2B1%20203-998-0403>
>>>>
>>>> * Listen to the Enterprise Java Newscast: *http://
>>>> <http://blogs.jsfcentral.com/JSFNewscast/>enterprisejavanews.com
>>>> <http://ww.enterprisejavanews.com>*
>>>>
>>>>
>>>>
>>>>
>>>
>>
>