incubator-adffaces-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Adam Winer" <awi...@gmail.com>
Subject Re: using trinidad as non-default render-kit
Date Wed, 28 Feb 2007 23:53:10 GMT
Changing how we get the RenderKit would break a lot of code,
so I'd be a big -1 on that.

There's other parts of the ExtendedRenderKitService
contract, and, yes, they'll be hard to enforce in general.
Facelets could actually fix this quite elegantly by
getting the renderKitId correct in createView(), which
should theoretically be possible.

I think the thing to do for this issue is to let CoreRenderer
call encodeBegin() if the RenderingContext hasn't been
called, and allow ViewHandlerImpl to recompute the
ExtendedRenderKitService after renderView() (since the
renderKitId will be correct by then).

-- Adam


On 2/28/07, Stefan Podkowinski <spodxx@gmail.com> wrote:
>
> This is definitely harder than I expected. What about coupling the
> RenderingContext with the used RenderingKit? And droping the
> RenderingContext ThreadLocal based access in favour of
> FacesContext.getCurrentInstance().getRenderKit().getRenderingContext().
> The RenderKitBase could probably extended for handling the
> RenderingContext?
>
> I dont quite understand the reason behind the
> ExtendedRenderKitService.encodeEnd() and
> encodeFinally() contract. The only implementation I found was the
> CoreRenderKit, and its doing nothing basically except for the
> RenderingContext creation. Theres also no such contract in the jsf
> RenderKit, so enforcement may turn out to become difficult.
>
> On 2/24/07, Adam Winer <awiner@gmail.com> wrote:
> > Stefan,
> >
> > Take a look at the code in ViewHandlerImpl - w/regard
> > to ExtendedRenderKitService and its encodeBegin() method.  You
> > don't need to have any direct reference to CoreRenderingContext.
> >
> > (Also, by only mucking with encodeBegin(), this'll fail
> > if the first Trinidad component is rendersChildren.  In
> > your page, perhaps it's not, but most actually are.)
> >
> > Once you use that, then you have to satisfy the contract
> > of ExtendedRenderKitService - calling
> > ExtendedRenderKitService.encodeEnd() and
> > encodeFinally().  That second half is the bigger trick,
> > especially because you really have to do
> > ExtendedRenderKitService.encodeBegin(), etc. exactly
> > once per page, not multiple times.
> >
> > -- Adam
> >
> > On 2/22/07, Stefan Podkowinski <spodxx@gmail.com> wrote:
> > > Adam,
> > >
> > > I think the CoreRenderer should take care of initializing the
> > > RenderingContext. Facelets will be kind enough to set the rendering
> > > kit in the FacesContext, as specified by the renderKitId attribute.
> > > Since encodeBegin() will be called in the CoreRenderer impl., we
> > > should know which RenderingContext we have to use, right?
> > >
> > > Try the following hack. Its still not 100% working, but the page is
> rendering.
> > >
> > > CoreRenderer.java:
> > >
> > >   public final void encodeBegin(FacesContext context,
> > >                           UIComponent component) throws IOException
> > >   {
> > >     if (!getRendersChildren())
> > >     {
> > >       RenderingContext arc = RenderingContext.getCurrentInstance();
> > >       if (arc == null) {
> > >         //throw new IllegalStateException("No RenderingContext");
> > >
> > >         try {
> > >            Class ctxClazz =
> > > Class.forName("
> org.apache.myfaces.trinidadinternal.renderkit.core.CoreRenderingContext");
> > >           arc = (RenderingContext) ctxClazz.newInstance();
> > >         } catch (Exception e) {
> > >           throw new IllegalStateException(e);
> > >         }
> > >
> > >       }
> > >
> > >       FacesBean bean = getFacesBean(component);
> > >       encodeBegin(context, arc, component, bean);
> > >     }
> > >   }
> > >
> > >
> > > On 2/22/07, Adam Winer <awiner@gmail.com> wrote:
> > > > How are you thinking of tackling this one?
> > > > I didn't have any great ideas.
> > > >
> > > > -- Adam
> > > >
> > > >
> > > > On 2/21/07, Stefan Podkowinski <spodxx@gmail.com> wrote:
> > > > > Created the issue in jira.
> > > > > https://issues.apache.org/jira/browse/ADFFACES-387
> > > > > I'll hopefully be able to contribute a patch in the coming week,
> if I
> > > > > have the time for it.
> > > > >
> > > > > On 2/20/07, Adam Winer <awiner@gmail.com> wrote:
> > > > > > Yep, re-reading it, that's exactly what you said.  Sorry
> > > > > > for the misunderstanding.  So this is a tougher nut
> > > > > > to crack.  I can imagine some hacks to try to
> > > > > > instantiate the rendering context on the fly, but
> > > > > > nothing really obvious and bulletproof comes to mind.
> > > > > >
> > > > > > I think we need a JIRA issue...
> > > > > >
> > > > > > -- Adam
> > > > > >
> > > > > >
> > > > > > On 2/20/07, Stefan Podkowinski <spodxx@gmail.com> wrote:
> > > > > > > Hi Adam
> > > > > > >
> > > > > > > I already have trinidad working with facelets. The problem
is
> it only
> > > > > > > runs with trinidad as the *default rendering kit*. Defining
> trinidad
> > > > > > > as the rendering kit per page does *not* work. Please see
my
> first
> > > > > > > mail for details.
> > > > > > >
> > > > > > > On 2/19/07, Adam Winer <awiner@gmail.com> wrote:
> > > > > > > > Oh, I see.  You're using Facelets.  You need to configure
> things
> > > > > > > > a bit differently than without Facelets.  Check out:
> > > > > > > >
> > > > > > > > http://wiki.apache.org/myfaces/Facelets_with_Trinidad
> > > > > > > >
> > > > > > > > -- Adam
> > > > > > > >
> > > > > > > >
> > > > > > > > On 2/16/07, Stefan Podkowinski <spodxx@gmail.com>
wrote:
> > > > > > > > >
> > > > > > > > > I just tried the other way around by setting
trinidad as
> default
> > > > > > > > > render kit in faces-config.xml and <f:view
..
> > > > > > > > > renderKitId="HTML_BASIC">. Afterwards the
page renders
> fine but I'm
> > > > > > > > > getting an error on submit:
> > > > > > > > >
> > > > > > > > > javax.faces.application.ViewExpiredException:
> > > > > > > > > viewId:/jsfexamples/examples-simple-1.jsf - View
> > > > > > > > > /jsfexamples/examples-simple-1.jsf could not
be restored.
> > > > > > > > >         com.sun.faces.lifecycle.RestoreViewPhase.execute(
> > > > > > > > > RestoreViewPhase.java:180)
> > > > > > > > >         com.sun.faces.lifecycle.LifecycleImpl.phase(
> LifecycleImpl.java
> > > > > > > > > :248)
> > > > > > > > >
> > > > > > > > > If I remove trinidad as default rendere kit,
submit works
> too.
> > > > > > > > >
> > > > > > > > > Whats also a bit concerning here is that with
trinidad as
> default
> > > > > > > > > render kit, the following hidden field is also
generated
> every time
> > > > > > > > > for the form tag, not taking <f:view renderKitId="..">
> into account:
> > > > > > > > > <input type="hidden" name="javax.faces.RenderKitId"
> > > > > > > > > value="org.apache.myfaces.trinidad.core" />
> > > > > > > > >
> > > > > > > > > On 2/16/07, Adam Winer <awiner@gmail.com>
wrote:
> > > > > > > > > > Stefan,
> > > > > > > > > >
> > > > > > > > > > We have a JSF 1.2 branch of Trinidad which
is well
> tested,
> > > > > > > > > > and contains (nearly) the latest code.
> > > > > > > > > >
> > > > > > > > > >
> > > > > > > > >
> http://svn.apache.org/repos/asf/incubator/adffaces/branches/faces-1_2-070201/
> > > > > > > > > >
> > > > > > > > > > The only bad news is that you currently
have to build
> > > > > > > > > > it yourself - we don't have an automated
build going for
> > > > > > > > > > this branch.
> > > > > > > > > >
> > > > > > > > > > (FYI, we rebranch every once in awhile,
and the URL
> changes
> > > > > > > > > > when we do.)
> > > > > > > > > >
> > > > > > > > > > -- Adam
> > > > > > > > > >
> > > > > > > > > >
> > > > > > > > > > On 2/15/07, Stefan Podkowinski <spodxx@gmail.com>
wrote:
> > > > > > > > > > > Hello
> > > > > > > > > > >
> > > > > > > > > > > Currently it does not seem to be possible
to use
> trinidad with jsf1.2
> > > > > > > > > > > ri and facelets and not having trinidad
declared as
> default rendering
> > > > > > > > > > > kit.
> > > > > > > > > > >
> > > > > > > > > > > Removing
> > > > > > > > > > >
> > > > > > > > > > > <default-render-kit-id>
> org.apache.myfaces.trinidad.core
> > > > > > > > > </default-render-kit-id>
> > > > > > > > > > >
> > > > > > > > > > > from faces-config.xml and adding the
render kit
> declaration to the
> > > > > > > > > view root
> > > > > > > > > > >
> > > > > > > > > > > <f:view ... renderKitId="
> org.apache.myfaces.trinidad.core">
> > > > > > > > > > >
> > > > > > > > > > > will break the view handling process
with the
> following error:
> > > > > > > > > > >
> > > > > > > > > > > java.lang.IllegalStateException: No
RenderingContext
> > > > > > > > > > >         at
> org.apache.myfaces.trinidad.render.CoreRenderer.encodeBegin
> > > > > > > > > (CoreRenderer.java:159)
> > > > > > > > > > >         at
> > > > > > > > >
> org.apache.myfaces.trinidad.component.UIXComponentBase.encodeBegin(
> > > > > > > > > UIXComponentBase.java:668)
> > > > > > > > > > >         at
> > > > > > > > >
> org.apache.myfaces.trinidad.component.UIXComponentBase.__encodeRecursive(
> > > > > > > > > UIXComponentBase.java:1209)
> > > > > > > > > > >         at
> > > > > > > > >
> org.apache.myfaces.trinidad.component.UIXComponentBase.encodeAll(
> > > > > > > > > UIXComponentBase.java:721)
> > > > > > > > > > >         at javax.faces.component.UIComponent.encodeAll
> (
> > > > > > > > > UIComponent.java:890)
> > > > > > > > > > >         at
> com.sun.facelets.FaceletViewHandler.renderView(
> > > > > > > > > FaceletViewHandler.java:571)
> > > > > > > > > > >         at
> > > > > > > > >
> org.apache.myfaces.trinidadinternal.application.ViewHandlerImpl.renderView
> > > > > > > > > (ViewHandlerImpl.java:182)
> > > > > > > > > > >         at
> com.sun.faces.lifecycle.RenderResponsePhase.execute(
> > > > > > > > > RenderResponsePhase.java:106)
> > > > > > > > > > >
> > > > > > > > > > >
> > > > > > > > > > >
> > > > > > > > > > > After doing some debuging, I came to
the conclusion
> that this must be
> > > > > > > > > > > due to the trinidad ViewHandlerImpl.renderView()
> method
> > > > > > > > > > > implementation. The problem is the
way the
> ViewHandlerImpl wraps the
> > > > > > > > > > > delegate, but needs to setup the RenderingContext
> before the delegate
> > > > > > > > > > > is executed. Creating the RenderingContext
is
> currently done by
> > > > > > > > > > > CoreRenderKit.encodeBegin(). But this
won't never
> happen in this case,
> > > > > > > > > > > because facelets did not parse the
page including the
> <f:view
> > > > > > > > > > > renderKitId..> element yet, so we
don't know which
> render kit to use
> > > > > > > > > > > before calling the facelets delegate!
> > > > > > > > > > >
> > > > > > > > > > > So here's whats happening in ViewHandlerImpl:
> > > > > > > > > > > 160: ExtendedRenderKitService cannot
be found because
> trinidad is not
> > > > > > > > > > > the default anymore
> > > > > > > > > > > 172: CoreRenderKit will not be called
to setup
> RenderingContext
> > > > > > > > > (thread local)
> > > > > > > > > > > 182: delegation to FaceletViewHandler
> > > > > > > > > > > afterwards
> > > > > > > > > > >
> org.apache.myfaces.trinidad.render.CoreRenderer.encodeBegin(
> > > > > > > > > CoreRenderer.java:159)
> > > > > > > > > > > fails on non initialized thread local.
> > > > > > > > > > >
> > > > > > > > > > > What do you think? Is there a chance
to reimplement
> > > > > > > > > > > ViewHandlerImpl.renderView() so this
problem could be
> resolved? Any
> > > > > > > > > > > possible side effects?
> > > > > > > > > > >
> > > > > > > > > > > Thanks,
> > > > > > > > > > > Stefan
> > > > > > > > > > >
> > > > > > > > > >
> > > > > > > > >
> > > > > > > >
> > > > > > >
> > > > > >
> > > > >
> > > >
> > >
> >
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message