The 1.0 Spec -- a few comments;

Mark Waks ([email protected])
Wed, 3 May 95 13:09:29 EDT


Gavin replies to me:
>> As far as I know, we do *not* have this capability in VRML currently;
>> even if we add it, I suspect that the form may be different.
>
>Did you fall asleep before reading the end of the spec? This extensibility
>feature is in there near the end.

Hmm. No, I confess I hadn't actually read the spec in detail. (Since
I'm not implementing, I hadn't bothered. My error.) I am rather
surprised here -- I don't recall ever discussing extensibility when we
were talking about the thing, and I will admit to some reservations at
this point; I don't think we've discussed the infrastructure enough
for it to work well. (And I think that it's mildly inappropriate for
at least some of the schemes we might use for extending VRML, although
probably not horribly so.) But what's done is done; I'm not going to
argue the point right now.

I now *have* read through the spec, albeit quickly. No major surprises,
aside from the extensibility section. A few notes, though:

First, I notice that it doesn't include any of the changes we've
discussed in recent weeks. (The user-sensible name field on WWWAnchor,
and the replacement/supplement of LevelOfDetail with LOD.) Did we
decide to go with those for 1.0? If so, someone ought to update the
spec accordingly. Similarly, it lists the MIME type as "world/vrml";
I had thought that we had eventually decided it had to be "x-world/x-vrml"
for now. (Although I wasn't following that discussion closely, so could
be misremembering.)

On the subject of extensibility, I *do* want to emphasize the
suggestion I made a few weeks ago; at the time it was an idle point,
but if this sort of extension *is* built into the language, it becomes
more important. I'd like to suggest a convention that
experimental/extension node names be prefixed with "x-", a la MIME,
for essentially the same reasons. It will help avoid the kind of
confusion we've just seen, where someone sees a node and can't find it
in the spec. More importantly, it will help us be clear about what's
experimental and what isn't when we start moving some of these
experiments into the real spec (and, in all likelihood, mess around
with them while doing so).

This isn't a hard-and-fast rule; it isn't intended for the spec, or
even for browser implementation. It's just a convention, that I think
will make our lives a little easier in the long run...

Finally, an unrelated question: when *are* the browsers and tools
coming out? Those of us without the connections to get betas really
are hungering for the suckers, and we're now getting to around the
time that they were supposed to be coming out. (Personally, I've got
more than a few people on my back asking for pointers, and I have to
keep putting them off...)

-- Justin
Who really would like to start laying out
his virtual office...

Random Quote du Jour:

Virus! .signature mutant -- deformed into copy me a .signature your I'm hi.