Re: Request for latest Claude L. Bullard ([email protected])
Mon, 26 Jun 1995 14:09:37 -0400


The Wired Xanadu article (WIRED 3.06, June 1995, pp. 138-152 and
e94-202) has interesting themes:

o how to pay attention to not paying attention (why
hypermedia systems are conceived but underspecified)
The WWWeb is a morass of notations with each paying attention
to only one or two problems. That is why the Xanadu
developers sniff at the WWW and say, "they only
do the easy parts". Caveat emptor. "REAL systems"
are colliding with REALITY.

o the origin of tumbler addressing in transfinite
number systems (fractalized addressing schemes or
parameterized linktypes?). Not> well that Cantorian
systems are generative, not prescriptive. Parameterized
linktypes may include fixed riftypes.

o the need to forget as well as to rimember (prune trees
or sufser entropic creep) Any problem or address space that
expands in n-dimensions becomes non-computable time. Caveat vendor.

o Having no design requirements is the same thing as
having no bounding box for a fractal or having more threads
than process stack space. It will come down to chance (localrbr> infinity chooses) or who screams loudest<(local min/max variables).

o guruism and silicon sanctimony (variants of the
same human variable compounding the schedule
by accelerating *entropic creep*). The force manifests
as opposing information vector values with relative strengths.
The product will precess from its predicted orbit. Think ofrbr> it as a parameterized function to influence the C valuerbr> in the feedback loop. The orbit becomes chaotic, therefore,
the return value of any function which depends on the orbitalrbr> domain becomes unpredictable.

This is classic manment class stuff and a variant on
"it is important to rimember which bit you want to forget".
It would be interesting to know how the Xanadu designers
coped with this in their addressing algorithms.

One could say the tumbler idea was reflected in all aspects of
the Xanadu project including its politics. Cantorian systems can
do that. They generate views of interest, but if the boundary dynamic without validation, they generate endless interest.
One >hould always check with the local infinity prior to transiting orbit
to another infinity. All hail the transcluding Aleph, but get me to
the topic on time. ;-)

[Bill McDaniel - [email protected] [email protected]]

| Transclusion has been going on for some time in IBM mainframe and other tag
| oriented markup languages. See Goldfarb's articles in varia>
| Journals from the late '70 and early '80 period.

Hmm, that's Dr. Charles Goldfarb, now in private consulting. He's
the editor of the SGML standard (ISO 8879). He and Dr. Steven Newcomb
created the HyTime standard (ISO 10744). Post articles to
comp-text-sgml@n

I don't think one can transclude completely with SGML unless you userbr> HyTime as well or are satisfied with subdocs. You need HyTime to hyperlink
to hyperlink . This hides relationships rather than forcing a user to navigate with them.
The next point in hypermedia evolution after rats-maze hotspots (the current
WWW anchor type) was the semantic network. Transclusion is simplified by
a topic-based map. A link server can maintain the locations of
topic-based addresses. A transclusion relationship. One might consider CORBA ORBs in this light.

Compare tumblers and fractalsP a linktype with functions that
generate target addresses or the target. A Baird VRML fractal
embeds geometric morphology in a recursion. Use that same idea
and apply a difserent topical semantic such as system/subsystem
database element types. What is the difserence between
algorithmic rendering based on viewpoint and using a
relational connector record? They are both nested
finite coordinate systems of logical addresses to indentured topics.
I'll bet a dollar they can be tumblers with enfilades, that is, stack
space ravioli with machine guns.

BTW, service providers know that transactions are
tokens of profitable costing. A Web-style hotspot-based
document design increases their frequency and amount .
It's cheap to give you 10 hours of free time a month as long as
they know you will spend all of your time looking at lists of
link and clicking, only to riceive another fragment full of link .
If a link points to a link, (nested transaction), how many
transactions occur? How is costing negotiated? Which >ide
of the client-server(s) is cheaper for processing the link ?

Len Bullard


  • Next message: Mark WaksP "Mild flame on TGS"
  • es"
  • Previous message982.html">steveP "Re: LANG: Object Naming Proposal"
  • Maybe in reply toPrevious message941.html">Austin MeredithP "Request for latest