[Opensim-dev] Considerations regarding long term development

[Toni Alatalo]

On Sep 22, 2011, at 4:37 PM, Per Mint wrote:
> yes, I only have concerns about the core OpenSim functionnality and entities so far.

There is one plan that was once made, and even kind of agreed, for a major refactor of the core entities. The SOG/SOP (SceneObjectGroup / SceneObjectPart) refactor. Adam worked on it for a while, and had funding for it from reX, but then unfortunately got busy with other things had didn't finish it (and didn't want any of the money then either, as dropped it).

Christa / Diva has asked about it a couple of times and shown some kind of interest, and long ago some of the Intel lab folks wanted a part of the functionality (ability to add arbitrary state / info to the scene entities). Christa was thinking it could be the 0.8 target -- next big rework after the ROBUST / service connectors biz was made in 0.7.

The plan is in a PDF in an attachment in a post to this list: 
Adam Frisby on Opensim-dev, Refactoring SceneObjectGroup - Introducing Components. 
http://lists.berlios.de/pipermail/opensim-dev/2009-December/008098.html

I refer to it in a recent article:
"An Entity-Component Model for Extensible Virtual Worlds"
https://github.com/realXtend/doc/blob/master/arch_article/simple.rst (search for 'adam')

Just for information, that this is one area where substantial change has sometimes been considered.

In fact, little after I first got to know about Opensim .. back in late 2007, the devs (including the original author M. Wright / Darren IIRC) noted how the scene code should be refactored.  But it is the heart of core (the sim code), and a couple huge reportedly intertangled classes that are not easy to touch -- refactoring would be considerable work (all those being signs telling that the refactor really would do good).

Given how things seem to go currently, however, there is perhaps not so much interest in that as what is there sort of works, and implements the SL model that people use Opensim for. The original goals were also about making a generic 3d / vw app platform, not necessarily similar to SL always, and there the extensibility enabled by such a component model would be more critical. 

It might make sense to target stabilization of current Opensim as some sort of 1.0, and leaving possible core refactors to later, but I have no idea how the devs see the maturity of others parts and the whole situation.

At reX we have that Entity-Component model implemented in Naali/Tundra, so have less pressure for Opensim changes (it was really critical for us to get it somewhere) -- and it was implemented in the modrex module for Opensim anyways (just the extensibility part, without the core refactor but just adding extra to it).

Just for some (historical) info of what was once thought and attempted in this area :)

> Per Mint.

~Toni

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://opensimulator.org/pipermail/opensim-dev/attachments/20110923/12211738/attachment-0001.html>