jsr344-experts@javaserverfaces-spec-public.java.net

[jsr344-experts] Re: [spec-763] Injection and JSF Managed Artifacts.

From: Bernd Müller <bernd.mueller_at_ostfalia.de>
Date: Thu, 14 Mar 2013 12:16:05 +0100

Hi,

I would recommend DOCUMENT because its a long expected Feature
at least for Converters and Validators.

At conferences and in blogs it was anounced, for example (best known)
http://jdevelopment.nl/jsf-22/

It is fine to restrict to request scope. This is what Seam 3 recommends,
too.
Seam 3 now moves to DeltaSpike and the faces module is announced
to "DeltaSpike and JSF 2.2", which means some people rely on
implemented by 2.2


Bernd



Am 11.03.2013 16:40, schrieb Edward Burns:
> Hello Volunteers,
>
> I'm back in the office today and will be working through the emails.
> However, I need to bring something up that Manfred brough to my
> attention.
>
> The design intent of spec-763 was to make it so all JSF managed
> artifacts are injectable using CDI and common annotations. After
> initial implementation work was completed, Manfred uncovered a difficult
> interaction between this feature and state saving. In short, it is not
> possible without significant more engineering work to make the feature
> work reliably for UIComponents and their attached objects. Some options
> include:
>
> OPTION: ROLL_BACK
>
> Do not require that CDI and common annotations work with UIComponents
> and attached objects. This means the following artifact types would be
> removed from table 5-3 in section 5.4.1.
>
> javax.faces.component.UIComponent
> javax.faces.component.behavior.Behavior
> javax.faces.convert.Converter
> javax.faces.validator.Validator
>
> OPTION: DOCUMENT
>
> Rather than roll back the changes, document that injection of things
> with a scope equal to or narrower than Request Scope will not work.
>
> After discussion with Manfred, I recommend ROLL_BACK until we figure out
> how to implement it correctly with respect to state saving.
>
> ACTION: Please let me know what you think of this.
>
> Ed
>