Aries-* and pax-cdi sample project in error

[Charlie Mordant <cm...@gmail.com>]

Hi guys,

I've got a real unworking sample this time!
I migrated with the latest and greatest OSGI-EE stack (Karaf 4.1).

So, I may be wrong in my component interactions, but i tried by all my
strength!
One week of holidays, so many mvn clean install 'cause I'm kind of an OSGi
fan and also workoholic...
So now, my project is able to:
* Use the latest and greatest pax-1.0.0-RC2: nice stuff, but kinda hard
migration, GNodet if you want to discuss, I'm here for you:
cmordant1@gmail.com
* Use config admin with deltaspike property: my module is kinda young, but
ossed and the only one which is able to do so. should it be an aries
module, a pax-cdi one, a deltaspike one?
* My jpa-daos module (on aries jpa 2, openjpa, jta +
maven-blueprint-plugin) fails one on three compilation: in my point of
view, Openjpa is the best candidate for OSGi-EE (I tried the other, but it
was far more unstable).
* I didn't succeed to run JSR 303 with CDI: I tried using the latest HV
version, making fragments (they're commited), making shaded modules, etc.
It succeeds with cxf, but fails with Cdi, or the other way... They's two
integration test which fails every times.

So, after more than two months trying to upgrade that kind of a -Java
+Osgi_EE project maintained for about 6 years (I had in mind to make a kind
of an Osgi jhipster generator), I'll try to move to a docker/swarm project,
because the wall is too higher: I could sell OSGi to a FR-Cac company
instead of a Spring uService stack (as I'm now one of their trusted ones)
if I was able to trust my stack and have some arguments (I personally I'm
persuaded that OSGi is the way to go, but sad to tell that the stack is far
from being dry).
So, I'll be available from every 7pm UTC+1 (Paris) for any question or
support on this project (if it can help the OSGi-EE community to improve
their bundles), my project is kind of simple to build (mvn install) and
easy to fail (wait for compilation: it will fail either on persistence,
either on cdi-validation).
I'll always encourage OSGi adoption, because IMO it's the only way to go,
but please keep it working right, reproducible (if someone prove me that my
code is wrong for a good reason, I'll make public apologies), and
understandable by the humans (JavaEE is already kinda hard for the mass,
and I'm convinced that OSGi-EE can provide modularity kinda free).

 Best regards, for ever,

Charlie