Archive for July 11th, 2008

Jul 11 2008

OpenSim holding the immersive middle ground?

While impatiently waiting for a local build of Mono to complete, I explored the new from Google Labs (the Mono build left some unused capacity in the XP+IE part of the lab ). It was fun to take keyboard knowledge of the SL client and guess the ways to zoom, dolly, pan, orbit, and dive around one’s avatar in Lively—and of course, find everything was there with googlish care.

I read a reminder (from a review of Wagner James Au’s book on early days of Linden Labs) that an original intent of that Linden crew was to build a representational and immersive model of real world.  And somewhere between rest and awakening grew a fresh recognition about OpenSim-type paraverses. They still occupy some application space not quite like, but spanning gaps in use that exist among Linden’s Agni grid, Google’s Lively, and Google Earth. A paraverse seems such a reasonable effort to pursue—for although it might seem a pedestrian app to describe, once it exists, its fidelity with real world should allow easier connections to all sorts of business, while offering all the creative possibilities that can derive from human-created worlds, like having both gravity and flying, having weather and having it the way you prefer it, and so on.

No responses yet

Jul 11 2008

A new OpenSim standalone grid SVN_5411 has lived

Published by under OpenSim

The config of a 64-bit Ubuntu 8.04 Heron system, with local builds of mono, libopenjpeg, and ODE was successfully stood up this evening. Visibility issues that needed to be overcome included my not comprehending that Ubuntu would happily configure multiple static IP with a default bridging, so that even with just one NIC wired, both would bind. Strike one for cloning Windows server hardware features (or not) in Ubuntu…

So with only one IP address configured, and only one network cable attached, the second issue was reachable. The system’s resolution of its name did not work for region configuration and it was necessary to use the public IP (which typically is not reachable from within a NAT-ted DMZ) for the system name in the region config .xml files.

All that worked out OK, and logins were possible, and a tour of the sim worked out until the easternmost reaches were approached, and did not rez. As it turns out, a few more ports need to be opened between the OpenSim box and the outside world. As it is, a 33-region standalone was up with nearly no prims, but seamless real-world terrain tonight. With a bit more of the ongoing cooperation with network administration, all 40 regions should be ready for content building by this weekend.

The server worked beautifully with four threads on the Xeons (it boasts total 27,000 bogomips), the 40 regions with terrain start up in under 15 seconds and use all four cores. Also, with the client rezzing out to 512 meters, it is dramatic to see three regions rezzing at once, which is one more than I’ve seen on a single Core2 Duo-based system. The fourth thread seems to retain more avatar mobility while those regions rez. It’s all very nice so far!

One response so far