Re: Internet-Draft: JSON Hypertext Application Language

 >> I'd remove
 >> the "self" link since it is redundant. The resource consumer already
 >> knows the resource URIs.
 >
 > That may be true for the root resource, but not for the embedded
 > resources. I think it's 'tidier' to include it for all resource
 > objects as it is consistent and lends itself better to situations
 > where resources are included as partials within other documents.
 >

Hi, and thanks for that RFC.

At http://stateless.co/hal_specification.html it is mentionned that 
"Content embedded within a Resource element MAY be a full, partial, 
summary, or incorrect representation of the content available at the 
target URI ..."

When discovering HAL I found this line very useful in understanding the 
power of embedded resources + self link combination. Maybe the RFC 
should re-use this phrase ?

A second thought, but it may be out of line : since
* all sub-resources are embedded via a rel value
* have a self link
* can have a partial content
what goal does the "_links" special attribute serves ? couldn't links be 
considered as "empty embedded resources" ?

For example, the following :

{
   "_links": {
     "self": { "href": "/orders" },
     "next": { "href": "/orders?page=2" },
   },
   "_embedded": {
     "order": { ... }
   }
}

Could as weel be translated into :

{
   "_self": { "href": "/orders" },
   "_embedded": {
     "next": {
         "_self" : { "href": "/orders?page=2" },
     },
     "order": {
         ...
     }
   }
}

It seems to me that replacing "_links": { "self" : {} }  by a simple 
"_self" : {} doesn't impair the possibility to reference the page 2 
under a "next" rel.

Thanks for your precisions
François