Yes, stored in session, however they are not overwritten on concurrent
requests because if a new scope is opened, and there are any scopes open, a
request query parameter is added:

https://github.com/seam/faces/blob/develop/impl/src/main/java/org/jboss/seam/faces/context/RenderScopedContext.java#L118

This has the benefit of avoiding the ugly parameter if only one window is
open at a time, or only one request is submitted at a time. I highly
recommend a strategy like this for window-scope as well if it's possible to
identify when you have more than one browser open while doing this...

~Lincoln

On Mon, Oct 15, 2012 at 12:24 PM, Mark Struberg <[email protected]> wrote:

> Thanks Lincoln, indeed that sounds very useful.
>
> Where do you store it between the requests? I guess in the Session, right?
> Thus 2 concurrent requests would overwrite each other, isn't?
>
> I think that's another neat example which should actually be
> per-browser-tab.
>
> LieGrue,
> strub
>
>
>
>
> ----- Original Message -----
> > From: "Lincoln Baxter, III" <[email protected]>
> > To: [email protected]
> > Cc:
> > Sent: Monday, October 15, 2012 5:58 PM
> > Subject: Re: [DISCUSS] JSF Scopes
> >
> > Just for clarification - @RenderScoped holds contextual objects until the
> > next "Render Response" phase is completed, at which point the scope is
> > cleared. IMO, much simpler than the flash, and much more intuitive -
> > perfect for holding Faces Messages or other "must show up once" type
> > values.
> >
> > ~Lincoln
> >
> > On Mon, Oct 15, 2012 at 7:55 AM, Gerhard Petracek <
> > [email protected]> wrote:
> >
> >>  @ #1:
> >>  not in case of a redirect. you can keep them by using a jsf-api
> manually,
> >>  but in this case they will be stored in the flash-scope... -> you have
> 2
> >>  disadvantages:
> >>   1) you have to do it manually
> >>   2) you have to wait for jsf 2.2 to get a flash-scope which works as
> >>  expected
> >>
> >>  @ #2:
> >>  you are talking about something completely different. the
> window-context
> >>  isn't a cdi context implementation.
> >>  it's a mixture of a manager for the current window and the
> > representation
> >>  of the client-window which contains different properties (or
> information
> >>  like faces-messages which need to survive a redirect) as well as
> different
> >>  scope-types.
> >>  please have a look at the link i provided with my first mail. there
> you see
> >>  that the window-scope is handled like any other scope bound to a
> window (it
> >>  just has a different strategy to expire and you can't use features like
> >>  groups).
> >>  -> only if you close the window-context (again: which is >not< the
> >>  window-scope), you close all scopes bound to a window and you drop the
> >>  information for/about this window (like the faces-messages).
> >>
> >>  regards,
> >>  gerhard
> >>
> >>
> >>
> >>  2012/10/15 Mark Struberg <[email protected]>
> >>
> >>  > I don't understand. The FacesMessages are held by the
> > FacesContext.
> >>  >
> >>  > If we talk about a WindowContext we always have to distinguish
> between
> >>  the
> >>  > list of active window contexts and the 1 which is active for the
> > current
> >>  > request.
> >>  >
> >>  > LieGrue,
> >>  > strub
> >>  >
> >>  >
> >>  >
> >>  >
> >>  > ----- Original Message -----
> >>  > > From: Gerhard Petracek <[email protected]>
> >>  > > To: [email protected]
> >>  > > Cc:
> >>  > > Sent: Monday, October 15, 2012 11:26 AM
> >>  > > Subject: Re: [DISCUSS] JSF Scopes
> >>  > >
> >>  > >t hat isn't correct. i thought about that as well, but some
> > people use
> >>  it
> >>  > > instead of std. cdi conversations to get rid of the
> > disadvantages.
> >>  > > -> if you would close (or restart) the window-scope and you
> > don't have
> >>  an
> >>  > > independent window-context, you would lose the information stored
> > for
> >>  the
> >>  > > current window (like faces-messages you need in case of a
> > redirect).
> >>  > >
> >>  > > regards,
> >>  > > gerhard
> >>  > >
> >>  > >
> >>  > >
> >>  > > 2012/10/15 Mark Struberg <[email protected]>
> >>  > >
> >>  > >>  Well, the Window Context is just the backing for
> > @WindowScoped. We
> >>  > should
> >>  > >>  try to not overengineer things.
> >>  > >>
> >>  > >>  All the Conversation scopes are based on the WindowContext.
> >>  > >>  We should definitely get @ViewAccessScoped as this is
> > something most
> >>  > users
> >>  > >>  really love.
> >>  > >>  But I'm sure there is also quite some interesting stuff
> > in Seam3 we
> >>  > > should
> >>  > >>  look at.Thanks Antoine so far for the start.  Anything
> > missing in his
> >>  > list?
> >>  > >>
> >>  > >>  LieGrue,
> >>  > >>  strub
> >>  > >>
> >>  > >>
> >>  > >>
> >>  > >>
> >>  > >>  ----- Original Message -----
> >>  > >>  > From: Gerhard Petracek
> > <[email protected]>
> >>  > >>  > To: [email protected]
> >>  > >>  > Cc:
> >>  > >>  > Sent: Monday, October 15, 2012 11:00 AM
> >>  > >>  > Subject: Re: [DISCUSS] JSF Scopes
> >>  > >>  >
> >>  > >>  > imo we should create:
> >>  > >>  > - a client-window adapter (as spi) for a 1:1 delegation
> > to the
> >>  > >>  > client-window-api of jsf 2.2+
> >>  > >>  > - a client-window implementation for jsf 2.0 and 2.1
> >>  > >>  > - a window-context similar to what we have in codi
> > (that isn't the
> >>  > >>  > window-scope - see [1])
> >>  > >>  > - scopes + a fine grained api to manage them (the
> > window-scope is
> >>  > also
> >>  > >>  just
> >>  > >>  > a kind of conversation which is very similar to std.
> > cdi
> >>  > > conversations).
> >>  > >>  >
> >>  > >>  > we could think about a more specialized spi per scope
> > (since we saw
> >>  > > that
> >>  > >>  > some edge-cases with the ViewAccessScoped required
> > internal
> >>  > > workarounds)
> >>  > >>  > and we could skip some SPIs for now.
> >>  > >>  >
> >>  > >>  > the first step is an agreement about the api, spi and
> > the basic
> >>  > > behavior.
> >>  > >>  >
> >>  > >>  > regards,
> >>  > >>  > gerhard
> >>  > >>  >
> >>  > >>  > [1]
> >>  > >>  >
> >>  > >>
> >>  > >
> >>  >
> >>
> >
> https://cwiki.apache.org/confluence/display/EXTCDI/JSF+Usage#JSFUsage-Scopes
> >>  > >>  >
> >>  > >>  >
> >>  > >>  >
> >>  > >>  > 2012/10/15 Mark Struberg <[email protected]>
> >>  > >>  >
> >>  > >>  >>  Hi folks!
> >>  > >>  >>
> >>  > >>  >>  I finally like to start working on JSF scopes for
> > DeltaSpike.
> >>  > >>  >>
> >>  > >>  >>  We already have the following 2 features
> > implemented and working
> >>  > >>  >>  (including unit tests):
> >>  > >>  >>
> >>  > >>  >>  * JFS @ViewScoped Context support
> >>  > >>  >>  * JSF-2-CDI scope mapping.
> >>  > >>  >>  * injecting typesafe JSF messages
> >>  > >>  >>
> >>  > >>  >>
> >>  > >>  >>  The next item on my list is the
> >>  > >>  >>
> >>  > >>  >>  * @WindowScoped
> >>  > >>  >>
> >>  > >>  >>
> >>  > >>  >>  This is kind of a Session per browser tab. Does
> > Seam3 provide a
> >>  > > similar
> >>  > >>  >>  mechanism? If not, I suggest taking a peak what we
> > do over in
> >>  > > CODI.
> >>  > >>  >>  There is quite some trickery necessary, but we
> > finally found a
> >>  > > solution
> >>  > >>  >>  which works pretty well [1]. This is also the base
> > of the
> >>  > > windowId
> >>  > >>  feature
> >>  > >>  >>  most probably coming with JSF-2.2 btw.
> >>  > >>  >>
> >>  > >>  >>
> >>  > >>  >>  Once we have the @WindowScoped support we can
> > build much more
> >>  > > fine
> >>  > >>  grained
> >>  > >>  >>  conversation stuff which is perfectly browser tab
> > aware based on
> >>  > > it.
> >>  > >>  >>
> >>  > >>  >>
> >>  > >>  >>  Wdyt?
> >>  > >>  >>
> >>  > >>  >>  Any cool features related to this to look at in
> > Seam3?
> >>  > >>  >>  Who is interested to help hacking this stuff?
> >>  > >>  >>
> >>  > >>  >>
> >>  > >>  >>  LieGrue,
> >>  > >>  >>  strub
> >>  > >>  >>
> >>  > >>  >>
> >>  > >>  >>
> >>  > >>  >>  [1]
> >>  > >>
> > https://cwiki.apache.org/confluence/display/EXTCDI/JSF+WindowHandler
> >>  > >>  >>
> >>  > >>  >>
> >>  > >>  >
> >>  > >>
> >>  > >
> >>  >
> >>
> >
> >
> >
> > --
> > Lincoln Baxter, III
> > http://ocpsoft.org
> > "Simpler is better."
> >
>



-- 
Lincoln Baxter, III
http://ocpsoft.org
"Simpler is better."

Reply via email to