- From: Andy Seaborne <andy@apache.org>
- Date: Sun, 28 Dec 2014 20:32:49 +0000
- To: public-rdf-comments@w3.org
On 28/12/14 19:43, Ruben Verborgh wrote: > Hi Gregg, > >> Is following redirection a problem for your test harness? > > The redirection itself does not pose a problem. > The problem is it makes the test incorrect—and that's a serious issue. > Below is the transaction that happens with cURL (truncated for clarity): > > $ curl -ivL 'http://www.w3.org/2013/TriGTests/trig-subm-27.trig' >> GET /2013/TriGTests/trig-subm-27.trig HTTP/1.1 > HTTP/1.1 301 Moved Permanently > Location: http://www.w3.org/2013/TrigTests/trig-subm-27.trig > >> GET /2013/TrigTests/trig-subm-27.trig HTTP/1.1 > HTTP/1.1 200 OK > Content-Type: application/trig > > < > # In-scope base URI is <http://www.w3.org/2013/TriGTests/trig-subm-27.trig> at this point RFC 3986: sec 5.1.3: [[ Note that if the retrieval was the result of a redirected request, the last URI used (i.e., the URI that resulted in the actual retrieval of the representation) is the base URI. ]] which makes sense to me because using the old URI, then redirected, or new URI directly gets the same base. Andy > > However, the only correct way to parse this document > is with <http://www.w3.org/2013/TrigTests/trig-subm-27.trig> as base IRI, > since this is the URL with which the document has been requested. > >> Probably the cleanest is to change the README and .nq result files. I'm concerned that changing this actually invalidates tests > > Another option is to change the direction of the redirect, > so that TrigTests => TriGTests instead of the other way round. > This should keep compatibility with all existing tests (if they handle redirects), > while providing the correct parsing of all documents. > > Best, > > Ruben >
Received on Sunday, 28 December 2014 20:33:19 UTC