Open source Second Life add-ons still months away

By Eric Reuters

(Corrected: In 11th paragraph, please read “libsecondlife.org,” not “libsecondlife.com.” A corrected version follows)

SECOND LIFE, June 14 (Reuters) - Support for open source add-ons to Second Life that were first touted in January are still months away, according to Linden Lab, frustrating some open source efforts to improve the viewer software and the user experience.

Hopes were high that open source add-ons to the viewer would help make Second Life easier to use, especially for new residents. A steep learning curve has led a large percentage of new users — as many as 90 percent, according to Linden Lab’s own estimates — to abandon the service within a few months of thier arrival in-world.

In January, Linden Lab released the source code for the viewer software, which displays Second Life on users’ computers. At the time the company said the release would lead to user interface improvements and new “skins” for the software.

“We desperately need better usability,” said Giff Constable (Second Life: Foresti Svarog), manager of the software practice at the Electric Sheep Company (ESC), one of the companies that started experimenting with changing the Viewer code.

Developers at the ESC have built a prototype overlay to the Viewer called “SheepSkin,” which they claim makes Second Life smoother and easier to use. The problem with coding the skin, or any other enhancement to the software, is that with each upgrade or mandatory patch to the Viewer introduced by Linden Lab, the skin needs to be reconfigured for the current release, Constable said.

“We really want a plug-in architecture,” Constable said, drawing an analogy to the plug-ins available for Mozilla’s Firefox web browser. Plug-ins would allow extra features to remain persistent even as Linden implements changes and upgrades to the core viewer engine.

But Rob Lanphier (Second Life: Rob Linden), director of open source development at Linden, said the company was unlikely to rebuild the viewer around a plug-in, or modular, architecture. “Linden Lab has always culturally been about making lots of incremental improvements that add up rather than trying to boil the ocean,” he said.

Lanphier said the work needed to make the viewer support plug-ins was potentially too difficult and too risky. “One of the big reasons why Netscape/Mozilla disappeared from the mainstream for so long is because the Mozilla developers spent a long time rebuilding the code base from the ground up as a modular code base,” he said. “The great news for the world is that they did that work, and we’re benefiting from it today in 2007. However, it wasn’t a great success for the shareholders of Netscape, Inc or AOL.”

Instead of plug-ins, Linden is considering an alternate solution called an “API layer,” which would allow third parties like ESC to add improvements to the viewer that are persistent through upgrades. Work still hasn’t begun on an API, but the subject is under active discussion at Linden, he said. “The time to get something out there is best measured in months,” Lanphier said.

Linden’s plan is to release design specs for the API to the open source community, gathering feedback on the proposal. Once the specifications have been finalized, Lanphier hopes the API’s eventual users will help code it. “However, it’s not design by consensus; just a more transparent way of working,” Lanphier added.

While plug-ins offer a degree of simplicity over an API, the difference might not be significant, said Jonathan Freedman (Second Life: Otakup0pe Neumann), project manager at libsecondlife.org, a clearinghouse for discussion about the Viewer code. “There is no need to rewrite the application,” Freedman said. “An API with all the good parts exposed could help.”

The potential upside of the API could be enormous, Freedman said. Among the new features he envisioned were changeable skins, new user interfaces, improved inventory management, and tighter integration between Second Life and social networking sites like Twitter or Facebook.

Freedman said the danger in an API is the potential for hackers to code changes to the viewer that allows them access to other residents’ information. “Security is a big concern any time someone’s intellectual property is in play,” Freedman said.

While Linden considers an API layer, most of the open source work being done with the viewer is implementing minor bug fixes and security improvements. Jason Giglio (Second Life: Gigs Taggart), estimates about 25 programmers including himself are actively working on minor fixes to the Viewer. Giglio authored a change designed to limit the number of incoming textures a resident could receive at one time, a trick griefers used to crash the software.

Giglio is glad he can help improve the Second Life experience for its users, even if he gets no compensation from Linden Lab. “Most open source projects have a company behind it that has editorial approval in the end,” Giglio said. “Linden Lab does seem committed to open source.”

(Disclosure: Reuters is a client of ESC.)