Re: a JSON-based RESTful protocol.

[Dan Haywood]

Hi Mark,
The viewer does use JAX-RS, so I think you slightly misunderstand... the 
discussion is about the format of one of the @PathParams ("oid") in the 
@Path [1]

@Path("/objects")
public class DomainObjectResourceImpl {
    ...
    @GET
    @Path("/{oid}")
    @Produces({ MediaType.APPLICATION_JSON })
    public String object(@PathParam("oid") final String oidStr) { ... }
    ...
}

So the question is, what is the format of the URL hits this resource...  
eg, any of:
* http://localhost:8080/objects/com.mycompany.myapp.Order|123
* http://localhost:8080/objects/ORD|123
* http://localhost:8080/objects/OID:123
* http://localhost:8080/objects/order:123

As I've said, the OID implementation is determined by the configured 
object store, specifically the OidGenerator.  In turn, this has an 
OidStringifier which converts to/from strings.  So, for those that 
really want to, this format can be customised if required.

Cheers
Dan

[1] 
https://svn.apache.org/repos/asf/incubator/isis/trunk/viewer/restful/viewer2/src/main/java/org/apache/isis/viewer/restful/viewer2/resources/objects/DomainObjectResourceImpl.java



On 25/06/2011 19:48, Mark Struberg wrote:
Agreed. Why not just use JAX-RS (JSR-311). This provides a @Path annotation 
where you can specify a different subdirectory in the URL.

LieGrue,
strub

--- On Sat, 6/25/11, Alexander Krasnukhin<the.malk...@gmail.com>  wrote:

From: Alexander Krasnukhin<the.malk...@gmail.com>
Subject: Re: a JSON-based RESTful protocol.
To: isis-dev@incubator.apache.org
Date: Saturday, June 25, 2011, 1:49 PM
I have a very bad bad feeling about
assuming all the class names are unique.

On Sat, Jun 25, 2011 at 11:58 AM, Robert Matthews<
rmatth...@nakedobjects.org>
wrote:

I've noticed a similar thing in a help index that I
have had to implement
recently.  While starting off with a set of fully
qualified class names I
noticed that for the domain classes (where the class
was ultimately accessed
by the user) we never used the same name as this would
be confusing to the
user. With that in mind we now assume that all the
class names are unique so
can safely refer to them using their short names.

Now as Dan said, these will depend on the OID in
question but we might want
to adapt its interface so it can provide both the
original internal form and
a suitable user identifiable form.  So in the
original example
com.mycompany.Order|123 would be available as
Order|123 and Dan's JPA
example, ORD|123 would also become Order|123.

Regards
Rob


On 24/06/11 19:38, Henry Saputra wrote:

Hi Dan,

Is there a way to map the actual Java class name
for the object such
as com.mycompany.Order|123 to short name? This
will help to make the
API URL simpler.

I am new with Apache Isis and naked object design
but looking forward
to implement it as prototype for our
architecture.

- Henry

On Tue, Jun 14, 2011 at 2:39 PM, Dan Haywood<dkhayw...@gmail.com>
wrote:
Alexander and I have (off-list) been throwing
an idea or two about
developing a viewer-agnostic network
protocol.  Alexander put a page on
the
wiki about this a while back [1].

My thought are that we could support this
using a JSON-based RESTful
protocol, perhaps as an extension/evolution of
our current Restful viewer
(which currently supports only XHTML).

With that in mind, I've added a number of
pages to the wiki to document
what
this protocol might look like; [2] is the
jumping off point, but see also
[3] for details on the typical representation
of a domain object.
Comments/contributions welcome!

Dan

[1]
https://cwiki.apache.org/**confluence/display/ISIS/**
Viewer-agnostic+Network+**Protocol<https://cwiki.apache.org/confluence/display/ISIS/Viewer-agnostic+Network+Protocol>
[2] https://cwiki.apache.org/**confluence/display/ISIS/**
Restful+Json+Protocol<https://cwiki.apache.org/confluence/display/ISIS/Restful+Json+Protocol>
[3] https://cwiki.apache.org/**confluence/display/ISIS/GET+**
on+an+Object<https://cwiki.apache.org/confluence/display/ISIS/GET+on+an+Object>





--
Regards,
Alexander