Any other feedback on this proposal ? When we discussed it originally
the EG seemed to be generally in favour of adding this kind of
functionality - is that still the case ?
Marc.
On Feb 19, 2008, at 12:46 PM, Marc Hadley wrote:
> I'd like to revive a thread started way back in Sept last year, see: https://jsr311.dev.java.net/servlets/ReadMsg?list=dev&msgNo=650
>
> The topic was support for URI-based content negotiation, essentially
> allowing a client to
>
> GET /foo.xml
>
> as an alternative to
>
> GET /foo
> Accept: application/xml
>
> I'd like to offer the following concrete proposal:
>
> - We only offer automatic support for content types, nothing for
> language or charset negotiation.
>
> - When the feature is enabled:
>
> * A request URI that ends with an extension is matched as if that
> extension were not present. E.g. @Path("widgets") would match
> requests for widgets, widgets.xml and widgets.json
>
> * If a URI template ends in a variable then the variable value is
> injected without the extension. E.g. @Path("widgets/{id}") with
> request for widgets/1.xml and @PathParam("id") String id would
> result in a value of "1" for id.
>
> * The extension is compared to the keys in
> ApplicationConfig.getExtensionMappings (Map<String, MediaType). If a
> match is found any Accept header value is replaced with the value
> for the matching key.
>
> - An @Path property style is provided to control the behavior. A
> value of 'platonic' means that the path should be treated as part of
> a platonic URI and the above behaviour is enabled. A value of
> 'distinct' means that the path is distinct and disables the above
> behaviour. The default value of 'default' defers to an application
> wide default specified as a property of ApplicationConfig (this will
> default to not enabled).
>
> - Existing UriInfo methods are not affected, any extension in the
> URI is included in the path returned by any of the methods. For
> convenience we add UriInfo.getPathExtension() that returns the
> extension or null if there isn't one, and
> UriInfo.getPlatonicRequestUriBuilder which returns a URI builder for
> the request minus the extension. We also add a
> UriBuilder.extension(String) that adds the supplied extension to the
> current final path segment.
>
> An example:
>
> @Path("widgets")
> public class WidgetsResource {
>
> @Context UriInfo uris;
>
> @GET
> @ProduceMime({"application/xml", "application/json")
> public WidgetList getWidgets() {
> ....
> }
>
> @POST
> @ProduceMime({"application/xml", "application/json"})
> public Response addWidget(Widget input) {
> Widget w = createWidgetEntry(input);
> URI platonic = uris.getBaseUriBuilder()
> .path(WidgetResource.class)
> .build(w.getId());
> URI distinct = uris.getBaseUriBuilder()
> .path(WidgetResource.class)
> .extension(uris.getExtension())
> .build(w.getId());
> return Response.created(platonic)
> .contentLocation(distinct);
> }
> }
>
> @Path("widgets/{id}")
> public class WidgetResource {
> ...
> }
>
> Assume you have an app config that maps "xml" to application/xml and
> "json" to application/json. Also assume you have the required msg
> body readers and writers for XML and JSON.
>
> GET /widgets.xml will get you XML, GET /widgets.json will get you
> JSON, GET /widgets will get whichever matches your accept header
> most closely.
>
> You can POST XML or JSON to /widgets.xml but you'll always get back
> XML. Same for /widgets.json but you'll always get back JSON.
>
> If you POST to a distinct URI you'll get a platonic location and a
> distinct content location. If you post to a platonic you'll get
> platonic location and content location (the latter is unfortunate
> but a custom message body writer could patch the value once the
> format is known.
>
> Thoughts, comments ?
> Marc.
>
> ---
> Marc Hadley <marc.hadley at sun.com>
> CTO Office, Sun Microsystems.
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe_at_jsr311.dev.java.net
> For additional commands, e-mail: dev-help_at_jsr311.dev.java.net
>
---
Marc Hadley <marc.hadley at sun.com>
CTO Office, Sun Microsystems.