- From: Boris Zbarsky <bzbarsky@MIT.EDU>
- Date: Fri, 01 Jul 2011 23:04:36 -0400
- To: public-iri@w3.org
On 7/1/11 10:20 PM, Bjoern Hoehrmann wrote: > So, you have this DOM and some event like clicking a link. You then have > some specification that defines, given the DOM and given the event, then > something or other happens. And this specification says, for instance, > > The reference is X and the base reference is Y; retrieve X. OK. So in this case X is an absolute URI and Y is the base URI, right? That seems to not match the behavior of any browser, for what it's worth. Does it match the behavior of anything at all? Or are you saying that the "base URI" for the purpose of this section should be something other than the actual base URI of HTML documents (that thing used for relative URI resolution), if the HTML spec wants? Perhaps the problem is that none of the specification writers writing specifications that used URIs understood the distinction between the base URI that a reference was actually resolved with respect to and the base URI as it applies for the purposes of this section, with the result that they don't actually define what the base URI to be used for this section is? > So I don't see how RFC 3986 can be a problem here. It would help a lot > to have an example where a specification defines that you have a refer- > ence X, a base reference Y, where X is the same as Y minus the fragment, > where you want to retrieve X with a new retrieval action (but can't say, > for instance, that there is no base reference in force). If you're allowed to use different base references for this section and relative URI resolution then ipso facto there is no such situation; you can _always_ define Y to be the empty string for purposes of this section, for example. -Boris
Received on Saturday, 2 July 2011 03:05:04 UTC