[Opensim-dev] OGP and Region Handles (was: Re: [Gridnauts] Next step?)

[Suzy Deffeyes]

Hi Mana,

The use of the regionhandle is an area of the OGP that could use some 
sunshine and disinfectanct. The reason things seem a bit inflexible right 
now is that the viewer doesn't really have a concept of managing its 
region list on a *per grid* basis because up until now there was only one 
grid. 

Also, during the original proof of concept phase of OGP, we found the 
regionhandle computation to be an assumption that wasn't spelled out in 
OGP.  The  rez_avatar/place protocol passes back a region_x and region_y 
which the viewer uses to compute a regionhandle.... *and* this handle must 
match the regionhandle that the sim computes for itself.  The assumption 
was that both the viewer *and* the sim are computing the regionhandle, and 
unless they compute it in exactly the same way, the viewer and sim can't 
connect properly. 

Additionally the client uses the coordinates stashed in the regionhandle 
in a bunch of different ways ( find the regions next to you, some graphics 
transform code, audio, other ways i haven't figured out...).  That will 
take some sorting, and pondering about how regionhandles should be used in 
the protocol. 

So it's exactly issues like this that we need to work out, and make sure 
they make sense for anyone ( SL, OpenSim, or otherwise) that is 
implementing OGP. 
This
Thanks
Suzy Deffeyes
Virtual Worlds 
Digital Convergence EBO
IBM Research
512.838.8770
suzyq at us.ibm.com


gridnauts-bounces at lists.secondlife.com wrote on 08/22/2008 03:12:37 AM:

> Hi Gridnauts,
> 
> Teravus Ovares wrote:
> > No, you're correct here.   The main difference is what OpenSimulator 
> > and OGP considers one unit on the grid space.  It's clear that OGP 
> > uses meters as the unit in terms of grid space definition. 
> > OpenSimulator uses blocks of 256 meters as the unit in terms of grid 
> > space definition. 
> > 
> > This leads to space allocations on the aditi grid like the 
> > following:  1048832/1048832,
> > translating to the following grid space in OpenSimulator: 4097/4097
> > (1048832 divided by 256 equals 4097)
> > 
> > One thing to note.  When it gets to the client, it's defined in terms 
> > of meters, not blocks of 256 meters.  This means that the use of the 
> > 256 blocks as one grid unit is limited to the way OpenSimulator 
> > represents grid locations internally.
> > 
> > This seems like a topic to send to the opensim-dev list as well
> 
> Hmm, I wonder why this should be an issue at all. Isn't OGP supposed to 
> allow teleporting from grid to grid? So it would be dangerous, to rely 
> on position or region handle anywhere in the process.
> 
> It might easily happen, that you teleport from a region in one grid to 
> another region with the same position/handle in another grid. If this 
> scenario is supported by the protocol, then why would it be necessary to 

> adapt region positions to the same scheme?
> 
> Or did I get something fundamentally wrong?
> 
> Best regards,
> 
> Mana
> 
> _______________________________________________
> Click here to unsubscribe or manage your list subscription:
> https://lists.secondlife.com/cgi-bin/mailman/listinfo/gridnauts
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://opensimulator.org/pipermail/opensim-dev/attachments/20080822/1e17d159/attachment-0001.html>