[jsr342-experts] Re: CDI in Application Client Container

From: Jim Knutson <knutson_at_us.ibm.com>
Date: Fri, 14 Dec 2012 15:01:39 -0600

Bill Shannon <bill.shannon_at_oracle.com> wrote on 12/14/2012 01:54:03 PM:
> Yes, thanks Pete. This is relatively easy to do and will meet our needs
> for now.
> While I recognize that the ACC isn't widely used, we have no plans to
> it and adding CDI support emphasizes how important we think CDI is to the
> Java EE programming model.
> Pete Muir wrote on 12/13/12 05:59:
> > Bill and I have agreed to basic support for CDI in the ACC - just
> app scope and dependent scope for now.

I don't think it's a great idea to increase the work load on EE vendors
just to make a philosophical point. Delivering a full EE compliant product
is already a substantial undertaking, but adding additional burden to that
for no value is just fueling the bloated platform rhetoric.

If the ACC was a high value piece of function and the industry was
CDI support in it, then I'd have no problem supporting this. However, the
recent community survey summary never mentioned anything about the ACC at
all. The lack of complaints about the ACC not supporting CDI also indicate
that the industry doesn't care.

I think we will have another chance to rethink the ACC when modularity
arrives and we start looking at how dynamic use of platform function fits,
but for now I would rather we change the EE 7 requirements so the ACC does
not require CDI.

Jim Knutson
WebSphere Java EE Architect