Re: [ogpx] A blog post from the HTML5/Websocket wars worth reading

[Morgaine <[e-mail]>]

I'll add a third observation which I also take from Tim Bray's article:

Defining *world* semantics (which loosely equates to Tim's *object model*)
rather than merely defining *interop protocol* semantics will
unavoidably *reduce
interop*.  If we limit ourselves to defining well every aspect of the
protocol *and only the protocol*, this is enough to guarantee compatibility
for interop.  In contrast, adding details of how worlds should work reduces
the set of candidate worlds that can benefit from the interop protocol, and
worse, acts against evolution and progress in those worlds.

VWRAP is not a protocol to enshrine a particular form of world, but a
protocol to define a particular form of interop.  We often make the mistake
of referring to "SL-like worlds" in these discussions, when what we really
mean is "VWRAP-compatible worlds".  There is certainly no need to be SL-like
nor even SL-compatible when using VWRAP.  For example, VWRAP will
undoubtedly be used by leading-edge worlds with services and assets that
Linden Lab chooses not to handle, as a business choice.

Therefore "SL-like" is a very incorrect phrase here.  We mean
VWRAP-compatible.


Morgaine.





===============================

On Wed, Feb 17, 2010 at 6:03 PM, David W Levine <[e-mail]> wrote:

>
>
> Kari Lippert <[e-mail]> wrote on 02/17/2010 11:36:55 AM:
>
> > [image removed]
> >
> > Re: [ogpx] A blog post from the HTML5/Websocket wars worth reading
> >
> > Kari Lippert
> >
> > to:
> >
> > David W Levine, ogpx
> >
> > 02/17/2010 11:37 AM
> >
> > Oh my, thanks for sharing... yes, I see some similarities too... but
> > I think in the case of VWRAP, we are obliged to provide the protocol
> > and also the "object(s)" including the semantics and intended
> > interpretation. If it is rich enough, the future will be able to
> > happen from within. And it the past is any indication, through
> > thoughtful re-purposing of things well defined!
> >
> > Kari
> >
> >
> >
> > http://kjlippert.wordpress.com/
>
> > http://community.webshots.com/user/MissGnomer
>
> My personal take away is twofold:
>
> First, describing "what" is sent is the win, not how to process it. So, you
> get to say
> "this is a prim who's visual shape is X" not "In order to display this prim
> you must take
> and apply the following behavior. "This is an idempotent motion update" not
> "Take motion update 46
> and apply it to the following state machine"
>
> Second, we want layering which follows Tim's notions that we not get into
> the multiple complex framing
> thoughts, and hiding what we are doing. The side by side comparison of the
> two specification styles is
> illuminating.
>
> Other take away thoughts as well, but those were my first two big ones. (I
> also plead guilty to wanting
> to make people aware of all the websocket/hybi stuff)
>
>
> - David
> ~ Zha
>
> _______________________________________________
> ogpx mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/ogpx
>
>