<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
Sean Dague wrote:
<blockquote cite="mid:4A4C9293.7030202@gmail.com" type="cite">
<pre wrap="">Stefan Andersson wrote:
</pre>
<blockquote type="cite">
<pre wrap="">I would argue that opening up for (reasonable) subclassing is a very core ideal of opensim-as-an-API.
And on the subject; somewhere in 0.7 land, I would argue that we have to rethink the whole scene/SOG/SOP bit - right now, state and behavior is intermingled to the degree that we cannot do anything that requires reasonable decoupling of state and behavior. Not to mention if one were to start working with using these objects in a non-core context.
</pre>
</blockquote>
<pre wrap=""><!---->
While I firmly agree, I've looked at that problem a couple times before,
and it's big. At the end of the day OpenSim is basically that scene,
then lots of things bolted on to it.
So nothing so grandious from me, but just some simple experimentation of
if there is a good way to have some set of synthetic server side
objects coexist with the user generated objects in the same scene, and
not turn into total chaos. :)
</pre>
</blockquote>
<br>
I've done this before (the traffic simulation). I love synthetic
objects that extend SOG, and all applications I'm involved with use
them heavily. Their only weakness right now is not being able to cross
borders like the other ones, because the deserialization routine
hardcodes the class. We should fix that at some point.<br>
<br>
Crista<br>
<br>
</body>
</html>