Re: Request for comments: "Origin of Document Species" and "Trust Mgmt"

David McCusker (
Thu, 05 Feb 1998 16:08:44 -0700

I need to contribute more often, if only to alleviate any frustration
from working hard all the time. (Sometimes it seems I'm months away
from doing something fun. :-)

Adam Rifkin (I Find Karma) wrote:
> The two papers Rohit and I had accepted for the WWW7 conference
> need to be submitted camera-ready by the end of the week. If you
> send me comments on either of the papers, I'll try to incorporate
> them before we submit the papers.
> 1. "The Origin of (Document) Species"
[ big snip ]

I have no comments for changing the Origin paper, but I like the
material a lot. I especially like the following sentence from section
five, since it appears a summary of some of my distobj commentary. :-)

| Documents exist as artifacts of larger processes like purchasing,
| reporting, or software development, and these uses bind semantic
| <em>meaning</em> to parts of a document.

I'm also interested by the idea that use of Turing-complete languages
is evolving toward more use of declarative encodings. As a exploratory
idea, it would be fun to investigate programming syntaxes that are more
declarative, in an attempt to bring these extremes closer together.

Of course, I'm really only thinking about this because I feel like
working on my Mithril (Ag) language again, because it sounds more fun
than Netscape tasks and my ongoing pd IronDoc (Fe) database work. But
I'd enjoy discussing language design with anyone interested in working
out XML style programming syntaxes to be used as official alternatives
to more traditional terse coding syntaxes. I'd like to support multiple
input (and export) syntaxes, including XML, lisp, and C style encodings.

With respect to distributed objects, this is related to Mark Baker's
comments weeks ago about using XML as an alternative to IDL interfaces,
except we could go further and consider writing code in XML as well.
It seems plausible to suppose that highly self-descriptive code would
be amenable to automated conversion for re-use in many contexts. (This
might be source re-use instead of more conventional interface re-use.)

An integrated development environment could display terse code to folks
accomplished at digesting concise syntax, but the IDE could export all
knowledge it has about relationtionships in XML for automated analysis,
and the more verbose syntax might be more approachable to folks on the
web who are not experienced coders.

I guess this sounds less interesting to folks who see Java as the end of
all language innovation. But it's also applicable to Java development.
I'd be glad to devote some time about this with anyone who wants to
develop ideas, as long as the ideas are for public domain consumption.

David McCusker, context-for-binding monger, storage and language weenie
Values have meaning only against the context of a set of relationships.