[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: Personal reply to Edd Dumbill's XML Hack Article wrt W3C XML Schema
- From: Gavin Thomas Nicol <firstname.lastname@example.org>
- To: "Simon St.Laurent" <email@example.com>, XML DEV <firstname.lastname@example.org>
- Date: Tue, 13 Mar 2001 02:54:36 -0500
> I'm coming more and more to the opinion that the only semantics which
> matter in the end are the semantics seen by the recipient of the message.
This is what I have been saying for a long time now (in my flames
against people saying that Qnames/URI's are in some way globally
If you look at any natural form of communication, you end up with:
foo ---- states -----> bar -+
+------- questions ---------+
+------- clarifies ---------+
. . .
The point being that *understanding* depends upon *interpretation*,
and *interpretation* is a function of the receiver, who in turn is
often affected by local conditions. The same is true of any
communication system, including software. Local interpretation
isn't just a good idea, it's a natural law!
There's also another factor playing into this too: the law of
evolution, which states that anything highly adapted to it's
specific environment is likely to not survive changes to that
environment. Likewise, software that is architected along rigid
lines is likely to be expensive to adapt to a changed environment.
Tightly bound types systems/data formats have the same problem.
One of the goals behind my "XSL Beans" idea, is to introduce a
level of indirection between the serialized data, and the
application-specific interpretation of the data, thereby providing
a natural means for evolution *and* migration. This has a clear
value for migratory agents and agent-to-agent (B2B, P2P) communication
An interesting thing I did years ago was build a pure peer-to-peer
distributed hypermedia system. The peers communicated in a very
tightly constrained S-EXPR syntax (flip the parens and you have XML).
The peers, in a reasonably sized community, exhibited "message evolution",
where the messages/data often evolved from the structure of the
initial message. This was emergent behaviour resulting from the
architecture, and was somewhat surprising to see, but easily
understandable once observed.
Still, maybe my local interpretation of my experience is leading me
down the garden path...