Yes. That's why I searched for a way to work around this, and I found one.

I'm still inline with the Spec JavaDoc excerpt you posted - the id's
are still prepended with UNIQUE_ID_PREFIX. I wouldn't expect users to
rely on the fact that the ids usually end with a number after this
unique_id_prefix.

regards,

Martin

On 7/14/06, Jacob Hookom <[EMAIL PROTECTED]> wrote:
 From the JavaDoc (1.1):
"Generate an identifier for a component. The identifier will be prefixed
with UNIQUE_ID_PREFIX, and will be unique within this UIViewRoot."

But resetting it on every encode breaks this contract, because the
UIViewRoot and other UIComponents survived across multiple
encodes/requests.  So it was a bug that was shared between the RI and
MyFaces.



Martin Marinschek wrote:
> You knew that in 1.1 this wouldn't work, right?
>
> the createUniqueId counter would have been reset on every request,
> both in the RI and MyFaces.
>
> regards,
>
> Martin
>
> On 7/13/06, [EMAIL PROTECTED] <[EMAIL PROTECTED]> wrote:
>> I'm not completely sure how Ed has setup the JSF JSP tags, but
>> Facelets goes purely off of the expectation that the UIViewRoot will
>> always produce a new unique value.
>>
>> When we bring in concepts like JspIdConsumer, which Facelets has too
>> with TagIds-- but it's only internal to the [Tag]Handler and
>> shouldn't need to be used as identifiers on the resulting output.
>>
>> Basically, each invocation of the Tag, the page's [Jsp]Id will be
>> assigned.  The Tomcat 6 stuff I worked on does something like:
>>
>> [URI HASH] + [POS OF TAG] + [COUNTER UNIQUE TO THIS TAG]
>>
>> The reason for this is repeatability for mark'n'sweep processing.
>> We'd like to be able to compare set identifiers on the component
>> model for succeeding passes to know when a component is brought in
>> our out of the Component Tree while allowing for JSTL/Control tags in
>> the page.
>>
>> Facelets instead sets a unique Attribute on the UIComponent, leaving
>> the actual ID up to the UIViewRoot/Impl.
>>
>> -- Jacob
>>
>> >Yes, my qualm is with the 'jsp' prefix. My problem is that (even though
>> >it would be good if we didn't) we have many automated tests that have
>> >been written using the old ids (and they are the same ids as the RI, so
>> >the same tests work with the RI as well).
>> >
>> >I know the 'right' long-term solution is to make our tests
>> >id-independent (or assign ids to everything in out JSP file). Given
>> that
>> >this isn't a JSF 1.2 implementation, I was hoping for a short term fix
>> >to my problem, but it sounds like your current solution depends on the
>> >prefix being different from the Servlet id prefix.
>> >
>> >> -----Original Message-----
>> >> From: Jacob Hookom [mailto:[EMAIL PROTECTED]
>> >> Sent: Wednesday, July 12, 2006 11:51 PM
>> >> To: MyFaces Development
>> >> Subject: Re: Ids in 1.1.3 and 1.1.4
>> >>
>> >> Unique Id generation aside, I think the qualm was with the 'jsp'
>> >prefix
>> >> in the generated IDs.  I know that JSP 2.1 has the ability to assign
>> >ids
>> >> unique to a tag on a page, but that's different than what you are
>> >> describing.
>> >>
>> >> Martin Marinschek wrote:
>> >> > The problem is that there is a bug in the spec here.
>> >> >
>> >> > If you called createUniqueId in earlier MyFaces versions, you would
>> >> > receive a unique id, but only unique for one request. On the next
>> >> > request, the sequence would be restarted. This would make working
>> >with
>> >> > dynamic component creation exceptionally hard, if not
>> impossible. On
>> >> > subsequent requests, the sequence counter would'nt start anew, and
>> >> > you'd get DuplicateIdExceptions all over!
>> >> >
>> >> > To fix this, we went to be more inline with what JSF1.2 foresees
>> for
>> >> > id creation. JSF 1.2 wants the servlet container to create unique
>> >ids
>> >> > - so we loose control about the ids produced, and createUniqueId is
>> >> > never called. Note: when we all move to 1.2, you won't know about
>> >the
>> >> > id-syntax anymore anyways - the servlet container provides the
>> >unique
>> >> > ids then.
>> >> >
>> >> > We implemented a partial solution of this - we have a separate
>> >counter
>> >> > for JSP-tags; and those ids are produced out of a different
>> >sequence.
>> >> > To make sure that the id's don't match, we prepend a "jsp" in front
>> >of
>> >> > the actual unique id counter.
>> >> >
>> >> > With this solution, we can store the current sequence counter of
>> >> > createUniqueId between requests, and dynamic component creation is
>> >not
>> >> > a problem anymore!
>> >> >
>> >> > I wonder why you would need the component id to have a special
>> >syntax?
>> >> > Do you rely in any way on this?
>> >> >
>> >> > regards,
>> >> >
>> >> > Martin
>> >> >
>> >> > On 7/12/06, *Gary VanMatre* <[EMAIL PROTECTED]
>> >> > <mailto:[EMAIL PROTECTED]>> wrote:
>> >> >
>> >> >     >From: "Matthias Wessendorf" <[EMAIL PROTECTED]
>> >> >     <mailto:[EMAIL PROTECTED]>>
>> >> >
>> >> >     >
>> >> >     > Hey Howard,
>> >> >     >
>> >> >     > you mean stuff like ?
>> >> >     >
>> >> >     >
>> >> >
>> >> >
>> >> >     >
>> >> >
>> >> >
>> >> >     > Was introduced in " Revision 359449"
>> >> >     >
>> >> >     > see
>> >> >     > http://svn.apache.org/viewvc?view=rev&revision=359449
>> >> >     <http://svn.apache.org/viewvc?view=rev&revision=359449>
>> >> >     >
>> >> >     > I can't tell you much about that. sorry!
>> >> >     >
>> >> >
>> >> >
>> >> >
>> >> >     I did some research on this for Shale Clay.  Clay has to
>> >> >     conditionally decide to use the ViewRoot to create a unique id
>> >or
>> >> >     use another method.  It looks like it has to do with partial
>> >page
>> >> >     rendering.   I suspect it was added for ajax component support.
>> >> >
>> >> >     That's pretty much the same answer as Matthias.
>> >> >
>> >> >
>> >> >
>> >> >     > -Matt
>> >> >
>> >> >     Gary
>> >> >
>> >> >     >
>> >> >     > On 7/12/06, Abrams, Howard A wrote:
>> >> >     > >
>> >> >     > >
>> >> >     > >
>> >> >     > >
>> >> >     > > Hi,
>> >> >     > >
>> >> >     > >
>> >> >     > >
>> >> >     > > Is there any reason why the UIComponentTag appends "jsp" to
>> >all
>> >> >     the
>> >> >     > > component ids in 1.1.3/1.1.4? Any chance this can be
>> removed
>> >> >     for 1.1.4 so
>> >> >     > > that it returns the same ids as UIViewRoot.createUniqueId()
>> >and
>> >> >     the RI?
>> >> >     > >
>> >> >     > >
>> >> >     > >
>> >> >     > > Thanks,
>> >> >     > >
>> >> >     > > h
>> >> >     >
>> >> >     >
>> >> >     > --
>> >> >     > Matthias Wessendorf
>> >> >     >
>> >> >     > further stuff:
>> >> >     > blog: http://jroller.com/page/mwessendorf
>> >> >     > mail: mwessendorf-at-gmail-dot-com
>> >> >
>> >> >
>> >> >
>> >> >
>> >> > --
>> >> >
>> >> > http://www.irian.at
>> >> >
>> >> > Your JSF powerhouse -
>> >> > JSF Consulting, Development and
>> >> > Courses in English and German
>> >> >
>> >> > Professional Support for Apache MyFaces
>> >>
>> >>
>> >> --
>> >> --------------------------
>> >> Sent from my FrankenBerry Wireless Handheld
>> >>
>> >
>>
>
>


--
--------------------------
Sent from my FrankenBerry Wireless Handheld




--

http://www.irian.at

Your JSF powerhouse -
JSF Consulting, Development and
Courses in English and German

Professional Support for Apache MyFaces

Reply via email to