- From: <keshlam@us.ibm.com>
- Date: Thu, 22 Jun 2000 20:14:31 -0400
- To: "Henrik Frystyk Nielsen" <frystyk@microsoft.com>
- cc: "David Carlisle" <david@dcarlisle.demon.co.uk>, abrahams@acm.org, xml-uri@w3.org
Henrick, I suspect you missed the part of the discussion which a) Agreed that for purposes of interlinking a set of documents which is always moved as a set, relative syntax makes perfect sense; b) Agreed that for purposes of linking to something which may not be moved as a part of that set, relative syntax is a disaster; c) Pointed out that for many -- arguably most -- uses of namespaces, one or more instances of the namespace identity are hardcoded into an application (the XSLT namespace is typical), and effectively is NOT moved as part of the selfcontained set of documents. The fact that relative references are a good thing for some purposes does not make them a good thing for all purposes. Namespaces are a purpose where they are in most cases going to be a Very Bad Thing indeed. One can make them meaningful... but the meaning you're forced to assign them flies in the face of the original intent of namespaces to define a reliably recognizable mechanism for grouping names. A namespace is _NOT_ primarily a link. It's an identity. Unstable identity is generally a bad thing, on the Web or anywhere else. ______________________________________ Joe Kesselman / IBM Research
Received on Thursday, 22 June 2000 20:14:51 UTC