My view would be that ADF 11g is "good enough" for custom enterprise
apps for the time being and that both ADFc/task flows and templates
have worked around some of the limitations in JSF 1.2. There was a big
hit, UI-wise, in going from 10.1.3 to 11.1 but then there were big
benefits too - I'd need some convincing that JSF 2.0 would bring the
same.
One of the advantages for customers adopting Oracle's ADF ought to be
platform stability - there are lots of competing toolsets/frameworks
but I don't think "enterprise" customers have the time or money to
keep rebuilding applications. In my experience, typically an
application is built in response to an initial set of requirements
(and business case) and then has a relatively long life with often
fairly minor change requests. You only need to look at most Forms
applications out there to see that. I suppose it's different for ISVs
where there's a steady revenue stream and customer demand for the
latest features.
Of course in theory, being a metadata-driven framework it shouldn't
matter too much what's underneath (JSF 2.0, Flex, JavaFX etc), but in
practice I expect that level of abstraction is very difficult to
achieve.
The main comfort factor for me is that there is now a juggernaut
development built on 11g so, whatever happens with JSF 2.0, Oracle is
going to have to have a super-slick migration process and provide
support/enhancements to ADF 11g for some considerable time.
Simon