[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]
Subject: [topicmaps-comment] Anomalies About the Use of Xlink in XTM
I think there are several odd things - or at least things that are missing - that loosely relate to the use of xlink references. I'd like to see what other people think about this. To start off, consider a topic reference that refers to a document outside the current document, one on another server, say: <topicRef xlink:href='http://someserver/theMap.xtm#t-dog'/> This is, by specification, a simple type of xlink, and only one type of semantic is defined for a simple xlink. It means essentially the same as an html href, which is that the link target is to be retrieved when the link is activated. However, we don't have any topic map specification that says how to make use of this target if the link is in fact activated. It's true that we have topic merging rules, but are they supposed to apply to this kind of topic? If we are never supposed to activate the link, on the other hand, why use xlink at all? The actual target is specified in the form of a bare xpointer. This by rights ought to mean that the thing to be retrieved is the entire element having the specified id. Doesn't this imply that the topic map engine on the remote server should convert the object with that id - a topic, in this case - to xtm and return it? Is this a reasonable expectation? If such an xtm fragment is returned, it is likely to have references to other topics, etc., from the remote map, but those wouldn't be returned by the xlink request. How should the requesting topic map engine deal with that missing information? Is the "orthodox" notion that the entire remote map should be imported and merged? I don't think that is always what is desired - what if that map were huge, for example? And I'll bet that few if any existing topic map engines have a way to receive a request for the entire topic map and to respond to it. If the referenced topic were not imported, then even its name could not be known, let alone any more detailed information. This would make it very hard to work with or even to present to a user. Similar issues exist for the other types of references, including subject indicator references. What should happen when one of these xlinks is activated? How can a request be made for follow-up information such as the topic type of a role? If these kinds of matters are not worked out, interoperability will be hindered (because there won't be good ways to make full use of information in a remote map). Also, we will hardly be able to say that topics maps could form part of the Semantic Web (whatever that may turn out to be). Comments and suggestions? Cheers, Tom P
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]
Powered by eList eXpress LLC