• From: "Len Bullard" <cbullard@h...>
• To: "'Michael Kay'" <mike@s...>, "'Ben Trafford'" <ben@p...>, <xml-dev@l...>
• Date: Mon, 25 Sep 2006 05:04:28 -0500

Almost Michael.  After more bytes consumed than I thought possible, I
conclude the problem is not that one can't have an identifier be a pointer,
but that the relationship between those two is in the eye of the consuming
object that has the methods to implement them.

IOW:  an identifier is an identifier to an identifying process, and a
pointer is a pointer to something that points.  The sticking point is not
specifying a polymorphic handler.  Not being clear that we are actually
talking about code when we talk about "linking" makes the version 80/20
chasm a whole lot bigger.

len


From: Michael Kay [mailto:mike@s...] 

An ISBN is clearly an identifier, not a pointer. An HTML @href attribute is
clearly a pointer, not an identifier. We need to use identifiers in
conjuction with a mechanism than can enable us to locate the things they
identify, but without "hard coding" the pointers in places where we should
use identifiers. The attempt to make URIs serve as both identifiers and as
pointers seems to be an unmitigated disaster.

• References:
  • RE: Xlink Isn't Dead
    • From: "Michael Kay" <mike@s...>