FYI, Mario Ivankovits has reminded me that the tomahawk t:messages
component has some related functionality.
When a component has an associated label component, then it can modify
the message by replacing the component's id with the text of the label.
Or something like that; hopefully the t:messages component describes
this feature in more detail. So if you are willing to use t:messages
instead of h:messages this might be useful...
Regards, Simon
Simon Kitching schrieb:
> Hmm. I know myfaces had a bug where resource-bundles are not available
> during the postback phase.
>
> I wonder if that ever got fixed. If not, that might be the cause.
> Usually resource-bundles are not needed during postback; generation of
> text output is done during rendering. Except that *validation messages*
> are of course resolved during the postback phase.
>
> I suggest you search the myfaces JIRA system to see if you can find an
> existing issue related to resource-bundles and postback. If so, then
> that's the problem. As a workaround, you might try writing a
> PhaseListener that always stores your resource-bundle into the current
> request scope; that should be about 10 lines of code.
>
> Anyway even using a simple literal for the label is at least better than
> what you had before, right?
>
> Regards,
> Simon
>
> j.palleschitz schrieb:
>
>> thanks simon,
>>
>> the label-parameter looked like a solution.
>> but it cannot resolve my resource-bundles.
>>
>> if i set label="#{label.username}" the message start with ": ..."
>> when label="Username" it looks fine.
>>
>> so it´s not the solution i´ve looked for.
>>
>> i can´t believe that such a behavior is wanted from myfaces.
>>
>>
>>
>> Simon Kitching wrote:
>>
>>
>>> While testing something else, I took a moment to check how Sun's Mojarra
>>> handles required-messages. It also inserts the clientId (ie *with* the
>>> form/subview prefix). Therefore myfaces 1.2.x and mojarra currently work
>>> the same. I think it unlikely that MyFaces would accept a patch to
>>> change its default behaviour to something different from Mojarra, although
>>> (a) you could try; the decision is not mine to make, and
>>> (b) you could create and offer a patch to make the behaviour configurable
>>>
>>> Re the "label" attribute I mentioned below: it looks like JSF also added
>>> this optional attribute to customise the {0} parameter, eg
>>> <h:input id="someProgrammerId" label="someUserLabel" required="true"../>
>>> will generate whatever generic "required" message is configured, but
>>> with {0} set to "someUserLabel". This feature still requires setting
>>> this on each component though.
>>>
>>> Regards,
>>> Simon
>>>
>>>
>>>
>>
>>
>
>
>
