• From: "Len Bullard" <cbullard@h...>
• To: <xml-dev@l...>
• Date: Sat, 23 Sep 2006 12:19:59 -0500

This thread is going to the same places it has always gone.  The Petri dish
is becoming stale.

O Style languages vs linking languages  (CSS vs XSLT, a useless debate
really)

O Linking languages vs programming languages (a useful debate because it
gets rid of the myth of separation of data and process as a 100% achievable
goal in all applications for all time).

O What are the types and semantics of the data in a linking structure?
See last point.   This is a system requirements debate.  Keep in mind that a
system is a set of constraints over a set of operations and don't get lost
in moreMetaThanThouLand.

So a more fundamental question:  can you have a link before you have a
process to create a link? 

The answer is no.  A link is an output routed to an input.  Links are data
therefore, transient products of linking processes.  That is why SGML
emphasized NOTATION and because the original web designs were explicitly
systemic, worked hard to get rid of variations there (constraints).

Is it important that a link be tightly coupled to a behavior, or is consumer
behavior tightly coupled to producer intension?

The answer is it depends.  That is the chasm that will not close and cannot
be closed permanently.  So any specification or standard for linking will
have a squishy center which ensures that interoperability of implementation
is never 100%.  You can achieve an 80/20 deal but you can't maintain it
forever.  Thus noise.  And you live with it.  Choose, commit, and live to
choose again.

The competitive game is to become the chooser of choices and sustain them.
The fiduciary game is to own those and extract profit from them.  Thus
patents, IP, eigen-value locking, and so on.  To keep the game from settling
into a minima that is suboptimum for some set of players, you try to avoid
Nash equilibria with uneven rewards, thus you work toward an evolutionary
stable strategy that reallocates rewards commensurate with
changes/innovations that keep all players in a competitive position.  

For that to work, links have to be transitory.  Pick the format and
environment that enables that.

len

• Follow-Ups:
  • Re: Xlink Isn't Dead
    • From: Dave Pawson <davep@d...>
  • RE: Xlink Isn't Dead
    • From: Ben Trafford <ben@p...>

• References:
  • Re: Xlink Isn't Dead
    • From: Jirka Kosek <jirka@k...>