Re: [vwrap] one question

[Cristina Videira Lopes <[e-mail]>]

Morgaine wrote:
> But in VWRAP, it is immaterial what kind of client application runs 
> the client endpoint of the VWRAP protocols, so the phrase "virtual 
> worlds that use the web browser as the client" doesn't really make 
> much sense in our context.  The answer is *Yes* only because we 
> scratch our heads and then ignore the phrase as an unnecessary 
> condition.  Sure, why not? :-)))

OK. So there are "client endpoints of the VWRAP protocol". Does this 
mean that there are defined behaviors for a VWRAP client on those 
endpoints? In other words, if my viewer is in JavaScript, I have to make 
the JavaScript program do things in specific ways, and not others, in 
order to be able to interoperate in VWRAP?

> Admittedly, your student would probably need to do some rather 
> unnatural coding since the VW model is really quite distant from the 
> Web model, and Javascript in the browser runs sandboxed so it's an 
> interesting question how your client would be coaxed to talk to 
> various external services, for example to be able to see assets worn 
> by visitors from other worlds.  (Remember that VWRAP is not tied to 
> the SL model in which everything is proxied through the current sim, a 
> highly non-scalable arrangement.)
CORS addresses that issue (avoiding the jasonp trick).
But this exposes the point I'm trying to clarify: on the web browser, 
VWRAP seems to be *forcing* application developers to use CORS, instead 
of leaving that as an independent engineering decision of each 
application. Why?

>
> I think that answers your question from the VWRAP end, but I get the 
> feeling there's something missing still.  Perhaps I could pose a 
> question of my own to help the discussion:  Do you consider a "virtual 
> world that uses the web browser as the client" to be significantly 
> different to a virtual world that doesn't define the type of client?
No; I'm interested in virtual worlds that are Web applications -- no 
more, no less. But the VWRAP protocol seems to be defining a specific 
type of client, and hence, a specific way of writing the JavaScript 
program -- at least wrt the endpoints.

>
> I would hope your answer is "No", since otherwise it would suggest 
> that worlds are going to Balkanize by the clients they use, which of 
> course would help nobody, and interop would be compromised.

Agreed.