[Opensim-dev] SceneObjectGroup vs SceneObjectPart

Tue Jan 22 21:21:21 UTC 2008

Sure thing, that example was just pouring out of my fingers, the vector syntax came from X3D and I don't like it either.

I would like some really standard albeit compact way of representing vectors/quaternions; surely, there must be something like MathML that does that? I'm thinking the pragmatics of html while still retaining extensibility of xml.

Also, by creating some 'model' group/part/shape classes that's more occupied with transformation of data than transformation of state, these classes could provide PrimitiveBaseShapes for the protocol, as well as more API-and-Xml-friendly property accessors.

I can start working on a suggestion sometime next week, if you all don't just race me to it.

Best,
/Stefan

> Date: Tue, 22 Jan 2008 15:56:49 -0500> From: sean at dague.net> To: opensim-dev at lists.berlios.de> Subject: Re: [Opensim-dev] SceneObjectGroup vs SceneObjectPart> > On Tue, Jan 22, 2008 at 08:35:42PM +0100, Stefan Andersson wrote:> > > It would be good to see a proposal on how to change those objects into> something better. None of us are in love with them, but better> approaches haven't popped up yet.> > Probably because nobody has made any attempt at discussing and proposing changes; this mail was an attempt to get that ball rolling.> > > > There was some energy poured into analyzing how attachments were communicated; that would be a good start, if we could get a wiki page up just detailing what we know about scene entities.> > > > * Related, we need to revise the xml serialization scheme and the db> > * schemes. The xml scheme should be user-friendly to the point where> > * you should be confident to create and edit objects in notepad,> > * basically. This is not the case at the moment. Massive duplication,> > * weird bit values and non-intuitive value ranges are king at the> > * moment.> > Given the amount of data in a prim, editing in notepad always is going> to be a bad idea. :) I think we've got something like 40 fields that> have to be sensible for the prim to work.> > >From working extensibly with shapes and serialization, I know that this is not the case.> > > > For example, consider something like> > > > <Object position="128.0 128.0 23.0">> > <Rectangle extrusion="straight" scale="1.0 2.0 5.0" offset="root"/>> > <Circle extrusion="sphere" scale="1 1 1" offset="0 0 1"/>> > </Object>> > > > that doesn't look like a bad idea at all to me.> > That would be pretty sucky to parse though. We should really be> breaking out the position bits into seperate data so they can be> accessed directly, and so there is no ambiguity of X Y Z or Z Y X for> values.> > I do like the approach of having sane zero values for things so they can> be left out if they are defaults. I think that will make life a bit> easier for making future portability here.> > -Sean> > -- > __________________________________________________________________> > Sean Dague Mid-Hudson Valley> sean at dague dot net Linux Users Group> http://dague.net http://mhvlug.org> > There is no silver bullet. Plus, werewolves make better neighbors> than zombies, and they tend to keep the vampire population down.> __________________________________________________________________
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://opensimulator.org/pipermail/opensim-dev/attachments/20080122/4ee19c07/attachment-0001.html>