ADMIN: VRML List Goals

Mark D. Pesce (mpesce@netcom.com)
Sun, 12 Jun 1994 21:40:26 -0700 (PDT)


Goals of the VRML Mailing List

This list has been established for a single purpose; to develop an open,
platform-independent specification for a Virtual Reality Markup Language
(VRML). This language will be used in conjunction with WorldWideWeb to
provide an standard to which viewer and editor applications and tools must
be written.

We are extremely pragmatic; although it is often expedient to develop
standards away from public view, and then announce them, fully formed, by
fiat, this is not the way that the Web or HTML have developed. Given their
success, it seems reasonable to want to emulate them.

We are starting out with some prejudices.

First, we don't have any language prejudices; we are open to all proposals.
We consider ourselves as starting from a position of ignorance, so we won't
arbitrarily dismiss any proposal because we think we know better. The wider
the net we can cast, the more successful our search for knowledge is likely
to be.

We feel, very strongly, that a solution which is proprietary,
which is "owned" in any sense by a commercial organization,
is unacceptable, and introduces complications greater than the task
at hand. Therefore, as proposals are made and reviewed, remember
that private solutions have a very poor history of success on Internet.
Nevertheless, the strength of a commercial proposal might rely on what
it can contribute to our own specification; we needn't steal from anyone's
work in order to benefit from it.

Finally, we feel it is important to establish a baseline specification
in front of and with the support of the people most likely to be using it.
To benefit from the thousands of years of experience within the subscriber
base is one of the most compelling reasons for having a list at all.

We wish to propose a process which we feel will result in the development of
an effective, if bare-bones specification, by the first of August. The
process has the following parts:

1) Request For Information

Within the next day, a post will be made the list outlining the requirements
for information on _existing_ languages which can fulfill the needs of VRML.
We ask that you DO NOT MAIL TO THE LIST, but, rather, send mail directly to
Brian Behlendorf (brian@wired.com), who will take the RFI traffic, compile
it into a daily (or twice daily) digest, and post it to the list.

We will give this process two weeks, from beginning to end. By then, all
interested parties will have made their contributions to our search for
information. This is just a request for information on existing languages;
if we, as a group, decide to write our own, that process would occur after
evaluating existing alternatives. We would prefer to avoid re-inventing
wheels.

2) Languages Review

We will be implementing HyperMail and WIT technologies on www.wired.com as
quickly as we can. The responses from the Request for Information will be
integrated with these browsing tools, and we will then enter a process of
evaluation, as the relative merits of the various proposals are considered.
We expect that only a few subscribers will contribute to this portion of the
discussion, and that most will simply review it from the daily digest
postings on www-vrml-digest.

This is the time when the list traffic will be heaviest; very little can be
done about this. If you wish to be actively involved in the formation of
the specification, you have to expect to spend at least a few hours a week,
during the review process, reading documentation and email.

3) Specification Drafting

Finally, the review period complete, it should be possible to begin drafting
a very basic specification. The first pass of the specification will not
need to describe every possible condition within a virtual environment; it is
enough to provide a static interface between the Web and three dimensional
geometry. The first specification should concern itself only with the
description of objects (including lights), and the scenes into which they
are placed. An object has no other behavior than that it can serve as an
anchor to another data source in the Web. The existing VRML specification,
at the VRML Forum WWW page, covers most of this territory. It is not
necessary to extend the scope of VRML functionality significantly from what
has already been defined.

A specification document will be drawn up and submitted for review.

4) Testing

List members will begin to write VRML compilers, parsers, interpreters, and
so forth, and will, from that, begin to have an understanding of the
shortcomings of the initial design pass. This information will be used in
the draft of a subsequent VRML 2.0 specification, which will also handle
"active" environments, where objects can be both dynamic _and_
communicative.

This process can be completed in the 7 weeks until the beginning of August,
and there is good reason to be quick about it; we want to finish our our
VRML viewer so that we can release it into the public domain, and actually
have people begin benefiting from the fruits of our labors.

We understand that this is possibly a more restrictive methodology than some
of you may have expected. While an absolutely free discussion of ideas and
techniques is a good thing, it is beyond the capacity of the more than 500
subscribers to this list. Our goal is to establish common ground quickly,
and use that as a basis for future work, without becoming bogged down in
procedural or prejudicial environments. Nonetheless, we actively invite you
to organize yourselves, if there are specific areas of interest which you
have a desire to focus upon.

Thank you for your cooperation and support.

Mark Pesce (mpesce@netcom.com)
Brian Behlendorf (brian@wired.com)

-- 
|| * That's AL * WorldWideWeb ftp://ftp.netcom.com/pub/mpesce/pubs/index.html