On Thu, Aug 2, 2012 at 11:45 PM, James Green <james.mk.green_at_gmail.com>wrote:
> John,
>
> I wrapped up the contents of the constructor in a try/catch to
> specifically log what was happening but nothing was logged.
>
> In the end I commented out all the classes for binding and the constructor
> was finally invoked.
>
> Then I was forced to uncomment the classes until I found the constructor
> no longer called. As you can imagine, I was swearing at the screen by this
> time.
>
> Anyway I now have the classes loaded having removed a few unnecessary ones
> and fixed a few others.
>
> My next job will be to make this work beyond Arquillian testing, into a
> real deployment to Glassfish, and to JBoss.
>
> Now, before I sign off I want to return the root of the problem: the lack
> of a cross-container default JSON notation. If it was not for this lack of
> standardisation I would not be wasting my time this week. My boss doesn't
> get why I'm having to put this time at all, after-all "Java must do JSON by
> now". So, I've heard there's a "future spec" to address this point, what
> news is there of it? I am specifically wanting to ensure that my code today
> does not rot into the future.
>
>
Your simplest fix would probably just to enable use POJO mapping. I am not
quite sure why you are starting with "natural" notation; maybe some old
tutorials refer to it?
POJO mapping is the usual "just works" method, based on Jackson's data
binding functionality.
The on-going Java JSR will not address data-binding between POJOs and JSON
at all, so it is mostly irrelevant here, and you don't need to worry about
that. Not to mention that even when it is finalized, JAX-RS spec would have
to support it, then implementations (Jersey) etc, so any effect it may have
will take quite a while.
-+ Tatu +-