At 07:28 PM 02/01/2000 -0600, Bob Crispen wrote:
>Miriam English wrote:
>
>> * database(s) for shared object states, behaviors, and ownership
>
>Since this was a requirements document ;-) I hope you'll
>forgive me if I point out that this item isn't a "what",
>but a "how".
Heheh actually my list was an "anything which randomly bubbled to the
surface of my brain so that I can stimulate discussions again" list. I
figured "what" and "how" would sort themselves out later. :-)
>My reason for being so picky, apart from being peeved that
>absolutely nothing happened this weekend (apart from my
>giving my SO a kiss that started in one millennium and
>ended in the next),
Heheh -- What a good idea! I will have to try that next New Year's.
>... is that we may end up focusing on "stuff
>databases can do" rather than "stuff shared 3D worlds need",
>and we may end up forgetting something important or focusing
>on an aspect of the problem that isn't as important as another
>area.
Good point.
(I will ignore the fact that the millenium doesn't end till 2000 winds
up... oops! So much for ignoring it :)
>So I'd like to suggest the following as a replacement:
>
> * Persistent objects whose state (size, color, position,
> anything that can be changed in VRML), if changed during
> the session, will reload in the changed state. Everybody
> who sees (hears, etc.) the object sees it in the same state.
Much clearer. :-)
> * Objects which are neither persistent nor shared, and
> which are free to stay put or behave for the private
> amusement and benefit of the person who acts on them,
> but which under no circumstances tell anybody else
> about their state.
>
> Yes, this is the default now, but in our zeal to make
> objects persistent, let's not have all the walls and
> ceilings and shrubbery broadcasting their locations
> at 60 Hz when there's no way they can ever change.
Things such as HUDs and reminders fall into this category too -- they show
only for individuals and don't persist beyond that person's existence in
the world. I imagine a butler bot (or Cati's sleeping FAQ robot) that could
show things to a person that nobody else would see.
> * An agreement on how persistent objects (the first kind)
> will be permitted to be used and how their state will
> change. For instance, A logs on, moves a chair to the other
> side of the table, and exits. B was there when A moved the
> chair, and stays around after A has left, moves the chair
> across the room, exits. Now A and B log on. Where is the
> chair?
>
> One answer that's used at CoCi, which is probably very
> difficult to implement, is that only owners can change an
> object's state. Unless A sells the chair to B, only A can
> move it. This is the "ownership" you mentioned? Another
> possibility is that a "public" chair can be moved anywhere,
> and as you log on, it appears where it was last left.
>
> Those were only example possibilities. Lots of other policies
> are good; not having a policy (and therefore having two people
> see the same chair in two different places) is bad.
Yes, I was specifically thinking of "ownership" being who gets to move,
re-texture or delete an object. Ownership is very useful to prevent vandalism.
Group ownership would be useful too. Public ownership would be the group of
everybody. (Possible use of wildcards here jumps to mind.)
> * Repository for shared objects and behaviors.
This is a very cool idea. This would probably end up just happening anyway,
but I can't help feeling it would speed things up if this got pushed.
> * Dynamic ownership: A throws a ball to B; B catches it and
> throws it back to A. While A is holding the ball, B can't
> throw it. After B catches the ball, B can throw it, but A
> can't.
Sounds like a public object fits the bill there. Perhaps what you mean is
something like: I bring an object into the world (I own the object) I put a
texture on it that I know you will like, then give it to you (you now own
it). Nobody but you may now alter that object, unless you give it to
someone else or widen its ownership.
>Let me add another, if I may (and please forgive me if this has
>been added to VNET since the last time I was in a VNET world):
>
> * Owner-controlled avatar behavior. Right now, I can click on
> Miriam's av and make it smile if that behavior was put in the
> world in response to a TouchSensor. I can't think of a way
> to make my own av smile.
Yep, Stephen has added this. You can code behaviors into your avatar and
VNet will add buttons on your gui that make those things happen for all
viewers. See Stephen's site and Jeff's for info on how to do it. I just
tried unsuccessfully to get to Stephen's site so here is the link to the
same text (I think) on Jeff's:
http://ariadne.iz.net/~jeffs/vnet/avatarBehaviours.html
This is much better than my primitive dog's wagging tail, which happens if
someone clicks on (pats) my dog avatar, but only happens for the person who
clicked.
>And I personally don't recall any widespread agreement on the
>following:
>
> * Experiment like crazy (yes, I know this is a "how") to find
> out what kinds of culling are best in what situations.
>
>That is, I think some folks did a few experiments, but I don't
>recall any conclusions, apart from "we can get this to work" or
>"we can't make this work".
Do you mean like the segmentation of worlds? I believe Robert (robofly)
Bjarnason was working on something like this. I prefer use of LODs over
segmentation of the world, but where LODs are not used segmenting the world
into cells would bring performance back into the realms of the usable.
Or were you referring to something else Bob?
Cheers,
- Miriam
-----------------------------------------------------------------
http://werple.net.au/~miriam/
Virtual Reality Association (VRA)
Melbourne, Australia
http://www.vr.org.au/
