Catherine Linden (Catherine Smith) has published a second episode of the new Linden Lab podcast series, Inside Linden Lab.
This month's episode features Joe Miller, Linden Lab's VP of platform and technology development.
Smith is sounding a lot smoother and more relaxed this time around. Joe Miller introduces himself and talks about technical strategy for Q1 2008. Miller hints at the upcoming culmination of a number of projects that he feels, overall, will contribute to significant improvements in capacity and stability for the Second Life grid.
Obviously a large component of that is the new Havok physics engine, and being able to avoid situations where the sim has to restart itself to pull Havok 1's nuts out of the fire. Miller believes that the Havok 4 code will roll out broadly (beyond the 200 sims of the Havok 4 Early Adopter program) in early February.
Mono, which is following on the heels of Havok 4, is another key project for speed and reliability. The current scripting runtime engine isn't very efficient, and Mono is expected to provide a performance gain of more than two orders of magnitude, and Miller also hints at improved memory constraints, though we don't expect any new language features in the near-term.
Miller also covers viewer crashes, but identifies a large blocker as simply identifying the causes of them. "Frankly, the crash-reporter has not been as helpful to us as we would like," observed Miller. A new and improved crash-reporter has been included in the latest release candidate builds, and Miller says that valuable data has already been gathered from the new reporting system.
Windlight (which reworks, speeds up and stabilizes the viewer rendering architecture - plus also providing eye-candy for higher-end PC users) is expected to appear in a release candidate in a few weeks (in 1.19.0?). Likewise the Dazzle reskinning of the browser is due in a First Look viewer in, Miller predicts, approximately three weeks.
While an effort was obviously made to minimize swamping listeners in technical information, we kind of felt that the podcast wasn't nearly technical enough. Don't hold back next time, Joe!
