[Opensim-users] Region coordinate confusion (was Re: Tibetan Classroom OAR

Karen Palen karen_palen at yahoo.com
Fri Jan 29 15:31:04 UTC 2010


You are correct about the +-128 notation, I am used to that notation from some other software and I was the one confused. I apologize if that confusion has spread to others.

For the rest, you have outlined the theory of what should be the mapping between a megaregion and a "simple" region.

I have actually used Diva's workflow as intended and it works quite well. 

However any graphical mapping is fraught with unexpected "special cases" which appear as bizarre results to the user. 

Other than the specific scenario that Diva explored, transformations between megaregions and "simple" regions is still pretty much untested.

I have managed to cause database corruption by moving OAR files between "simple" regions and megaregions. Corruption bad enough to require rebuilding of the entire database! 

I have not taken the time to explore and analyse the causes of these problems yet, but clearly great caution is required!

Bottom line: this feature needs testing, it should not (yet) be used for a system that has active users!

Karen

--- On Fri, 1/29/10, James Stallings II <james.stallings at gmail.com> wrote:

> From: James Stallings II <james.stallings at gmail.com>
> Subject: Re: [Opensim-users] Tibetan Classroom OAR
> To: opensim-users at lists.berlios.de
> Date: Friday, January 29, 2010, 6:45 AM
> Namaste :)
> Just to clear up any confusion about oar files - they
> are simple compressed xml representations of the
> regions' content and the assets required to reproduce
> them. The xml embeds object coordinates; the coordinates are
> taken directly from the objects.
> 
> As a result, it is certainly possible to load
> oars of one size and configuration on opensim instances of a
> different size and configuration. Results are predictable
> but not necessarily desirable. Terrains may exhibit bizarre
> artifacting and objects may be 'off-world'  if the
> oar size exceeds the target region size on any given
> axis.
> 
> Which brings us to coordinates: there is no
> practical use of a notation like "+/-128" in this
> context. I've never seen it used before and it is
> clearly leading to some confusion in this instance.
> 
> In opensim, region coordinates (the coordinates
> used to position regions on the grid), are in meters. Region
> cells (a subunit in opensim who's distinction is not
> meaningful in secondlife), are 256 meters on a side
> (identical to a canonical secondlife region). All objects in
> canonical region will have x/y coordinates that satisfy the
> condition '255>coordinate>0'. These may be
> combined into megaregions.
> 
> When a megaregion is in use, all the objects on
> the megaregion will, under normal circumstances, be present
> in the root region's database tables. They will have x/y
> coordinates that satisfy the condition
> '(subregions_along_the_axis*255)>coordinate>0'.
> 
> Diva Cantu has a workflow for merging simple
> regions and their contents into a single megaregion. I have
> not experienced success in my attempts with this workflow;
> this doesnt mean it is not effective; it means my attempts
> with it haven't been. That said, I'd suggest making
> backups at critical points in the process to avoid loss of
> content.
> 
> I hope I have brought light into the shade with
> this post :)
> 
> Cheers!James
> -- 
> ===================================
> http://osgrid.org http://simhost.com
> 
> http://del.icio.us/SPQR
> http://twitter.com/jstallings2
> http://www.linkedin.com/pub/5/770/a49
> 
> 
> 
> 
> -----Inline Attachment Follows-----
> 
> _______________________________________________
> Opensim-users mailing list
> Opensim-users at lists.berlios.de
> https://lists.berlios.de/mailman/listinfo/opensim-users
> 


      



More information about the Opensim-users mailing list