- From: Julian F. Reschke <julian.reschke@gmx.de>
- Date: Wed, 11 Apr 2001 19:36:55 +0200
- To: "Jim Whitehead" <ejw@cse.ucsc.edu>, "WebDAV WG" <w3c-dist-auth@w3.org>
Let's try to approach it this way: A property is a tupel of (namespacename, localname, content). content can be (for instance) empty, a string, or mixed XML content. By definition, it can by anything that belongs to the property element's infoset (including anything below it). If "content" is not a simple string, it needs to be stored in an XML-friendly way. For instance, elements or attributes in the content might use namespace prefixes that have been defined outside that element. For instance: <proppatch xmlns="DAV:" xmlns:x="foo"><set> <prop> <bar><x:baz/></bar> </prop> </set></proppatch> If attributes of the property itself need to be recorded as well, that's extremely easy to do: just instead of storing the "content" of the element, you store the XML serialization of the element including it's attributes and children. Two other notes: a) It needs to be specified which parts of the Infoset are preserved (comments, processing instructions)? I'd say: all of them. b) Microsoft's webfolders already use the old XML data namespace to put data type information into properties, for instance <DAV:getlastmodified dt:type="isoDate.tz" />. Julian > -----Original Message----- > From: w3c-dist-auth-request@w3.org > [mailto:w3c-dist-auth-request@w3.org]On Behalf Of Jim Whitehead > Sent: Wednesday, April 11, 2001 7:13 PM > To: WebDAV WG > Subject: RE: Issue: PROP_ATTR > > > > > The real question is about attr1. I see that attribute > > > as part of the element that *names* a property, but it > > > isn't part of the property *value*. > > > > That's an interesting comment -- I'd say that everything from > the opening > > element <theprop... to the closing element </theprop> is the > property, and > > the outermost element is called its 'name'. Everything about > the property > > should be stored, including 'copying down' namespace declarations where > > required. > > Since persisting the attributes is going to be tricky enough, my > leaning is > to only allow namespace and xml:lang attributes in the property name tag. > > However, I do agree that servers MUST persistently store XML attribute > information found in the value of the property. I don't care how > they store > it -- the XML is just a marshalling format. I agree with the suggestion to > look at the XML Canonicalization Recommendation for ideas on how > to specify > the round-trip behavior <http://www.w3.org/TR/xml-c14n>. > > It's important to support XLink style linking > <http://www.w3.org/TR/2000/PR-xlink-20001220/>, and this is > *very* dependent > on the use of attributes. Storage of XLinks in DAV properties is a very > natural way to persistently attach link metadata to a resource -- > the major > roadblock holding this up at present is the fact that, right now, a client > does not have a guarantee that attribute information will be persistently > stored. I'm glad to see the sentiment of the working group > leaning in this > direction. > > - Jim >
Received on Wednesday, 11 April 2001 13:37:36 UTC