[
Lists Home |
Date Index |
Thread Index
]
If there are no shared semantics, what are the messages
in a SOAP payload used for? "Whatever you want to use them
for is the chronically right answer", but it's wrong when
the humans get involved and the point is to involve the
humans. It can also be wrong if there are meta-turtles
in the stack.
There is a first turtle. They own the PDA, they sit
in the cockpit, at the desktop, on the phone, pick up
stuff at the fax machine. If you don't find them,
you aren't at the top or bottom of the stack.
XML names don't name types. Turtles use XML names
to name types. Constantly.
DRM = Digital Rights Management | Data Reference Model | PCDATA
Choose one. If you choose what the sender chose, you are
communicating. If not, you are in an error condition given
the scope of the set of messages you are sending given the
assumption that all of those messages are about your choice.
The trick is how many messages it takes before you,
the sender, or any intermediary detects the error and
sends an error message. If you misinterpret that message,
you do it again at some unscheduled cost. If this happens
often, you may want to derive/apply a contract for communications.
Do that too deeply and your costs will rise without
good ROI. Do that too shallowly and your error rate
will not meet predictions or an error will occur that
has a noticeable and unpredicted scale of consequence.
Choosing types based on names has a cost and when done
badly or not quickly enough, the cost goes off the scale.
The historical cause for the sinking of the Titanic was
a meta-turtle who pocketed the iceberg warning rather
than passing it through to the operational turtle-in-charge
because it indicated a potential but not yet realized
reduction in capacity to meet meta-goals of sales based
on operational performance. The meta-turtle survived the
sinking where the operational turtle-in-charge and 1500
customer turtles did not. The meta-turtles have a way
of doing that.
And so it goes. The world is festering with meta-turtles.
len
From: Nathan Young [mailto:natyoung@cisco.com]
> The first turtle is the human
> who is the ultimate semantic processor
Wait, there's a first turtle? I thought the stack was turtles all the way
down to a person. Does it matter if the first turtle is a person? Aren't
all the turtles people who can choose turtle-hood???
I don't want to turn this into an "angels on the head of a pin" argument
(or maybe I do). But it could be argued that there's never any such thing
as a shared language. There are just languages that are shared enough for
certain purposes. Although we hone the degree to which our shared
vocabularies allow us to communicate about the real world and our ideas,
there's never a "buck stops here" moment where rules become reality and
the description of the thing can truly replace the thing itself...
We can have a limited degree of absolute authority when we are alone. All
of us are authorities on our own private languages. It's the principle of
"king of all we survey". And as long as we don't involve anyone else, we
can maintain absolute authority over the correct interpretation of our own
thoughts (just don't mind reality). Once we involve someone else the
rules change and we're back to turtles all the way down.
I'm probably mangling your metaphor, and I've almost certainly "lost it in
the philosophy of meaningfulness".
People invent vocabulary every time they try to communicate. It's what
communication IS. The fact that XML is a tool for representing
vocabularies binds it to some crazily human ambiguity.
What does "DRM" stand for? What does "WML" stand for?
|
