On Sep 8, 2008, at 12:21 PM, Reto Bachmann-Gmür wrote:
> Hi Paul,
>
> by default a provider is instantiated only once per application. Do
> implementation have to change this behavior for Providers with
> request-specific injections? Re-injecting the UriInfo on each
> request doesn't seem an option because of concurrency issues.
The injected instance can be a proxy that obtains the per-request
instance from a thread-local variable. When the per-request stuff is
instantiated it can be set on the relevant thread local variable.
Paul.
> Cheers,
> reto
>
> Paul Sandoz wrote:
>> Hi Reto,
>> You can inject a UriInfo instance onto a field of a reader/writer,
>> for example:
>> @Context UriInfo ui;
>> Paul.
>> On Sep 8, 2008, at 8:31 AM, Reto Bachmann-Gmür wrote:
>>> Hello
>>>
>>> I was writing a MessageBodyReader accepting RDF/XML and returning
>>> an RDF-Graph instance when I noticed that my client uses relative
>>> URI references:
>>>
>>>
>>> PUT /editor/dynamic/page1 HTTP/1.1
>>> Host: localhost:8282
>>> Keep-Alive: 300
>>> Connection: keep-alive
>>> Content-Type: application/rdf+xml
>>> Referer: http://localhost:8282/editor/static/disco.xhtml
>>> Content-Length: 311
>>>
>>> <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
>>> <rdf:Description rdf:about="page1">
>>> <infoBit xmlns="http://discobits.org/ontology#"
>>> rdf:parseType="Literal">bla bla bla bla </infoBit>
>>> <rdf:type rdf:resource="http://discobits.org/ontology#XHTMLInfoDiscoBit
>>> "/>
>>> </rdf:Description>
>>> </rdf:RDF>
>>>
>>> In the MessageBodyReader I cannot reconstruct the full URI of the
>>> subject of the two triples in the graph as I cannot access the
>>> fist line of the http-request.
>>>
>>> I think the problem may occur with any formats supporting relative
>>> URIs, for RDF things are particularly difficult as relative-uris
>>> are permitted only in the serialization format, a valid RDF-Graph
>>> never contains relative URI references.
>>>
>>> For MessageBodyWriter the situation is similar if the applications
>>> wants to deliver relative URIs in the serilaized form. The issues
>>> with writers is however less urgent, as it's usually acceptable to
>>> deliver absolute URIs.
>>>
>>> Should MessageBodyReader have access to a context-URI? Or to the
>>> full Request instead of just the entity headers?
>>>
>>> Reto
>>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: users-unsubscribe_at_jsr311.dev.java.net
>> For additional commands, e-mail: users-help_at_jsr311.dev.java.net
>
>