Intel committed to mobile Linux, despite core dump
Participation through project packing
Linux FCS Intel remains committed to its mobile Linux project, it has said, despite handing over supervision to the Linux Foundation.
The chip giant has pledged it will provide additional engineers and resources to refine Moblin, the project to build a ready-made Linux and hardware stack for a range of consumer devices.
Despite pulling in more of its own people, though, Intel not only believes it can attract more independent participation in Mobiln. It will also use external participation to help measure Moblin's success.
Initial goals for Moblin are to achieve a two-second boot time, for use in the world of in-car systems, and for developers to build tools for Moblin's Clutter-based interface framework.
Imad Sousou, director of Intel's open-source technology center, told the Linux Foundation's Collaboration Summit in San Francisco, California this week: "One measure of success over the next year is how many non-Intel people are maintainers, are participating heavily, are taking decisions on what were calling the steering committee.
"That's something I measure our success on. If we don't do that, there is a problem We are looking for participation."
As with any vendor-lead project moving to the community, Moblin faces a tough slog in attracting outside interest not just because of where it's come from, but because Intel's going add more staffers. Moblin maintainers today are predominantly from Intel, Sousou admitted.
Intel's also pitching Moblin into a tough market.
Moblin is a nice idea for Linux fans and pragmatists out there, because it promises a prepackaged version of Linux and open-source software with a hardware stack for use in mobile devices. The idea is to provide a consistent experience across different devices.
That means consistency in things like the interface, programming framework and tools, wireless connection, and start-up times.
The target consumer companies that might adopt Moblin, though, will need more persuading.
Many are used to rolling their own hardware and Linux stacks for things like mobile phones and in-car navigation systems. Adoption will also be slowed by the need for these companies to accept and evaluate Intel's Atom and chips, which Moblin's tied to.
Changing software is one thing, replacing the hardware is a fundamental strategic decision that also has ramifications on overall design.
Under the auspices of Intel, Moblin hadn't done so well when it comes to convincing adopters. There was a major re-write mid-way through from Ubuntu to Fedora, which would not have generated much confidence, while OEMs would have had a problem with the fact Intel was running the project and using Intel hardware.
One of Linux's selling points in devices is that, unlike Windows, you are not tied to a single vendor.
As for developers, while Intel might actively encourage open-source it remains - like Sun Microsystems - a hardware company. That combined with Intel's corporate bureaucracy has failed to excited developers. The Fedora switcheroo will not have convinced people this has been the best executed or projects, either.
To be fair to the Moblin team, Reg sources have told us the project was pushed to deliver a first product before it was ready.
Sousou told Linux FCS: "Big corporations are not good shepherds of open-source projects." He noted Intel learned a great deal from version one.
What ever the reason, version one lacked integration and didn't work on netbooks - a major strategic blunder. The decision was therefore made to make a clean break with version two, which was released earlier this year, rather than evolve version one.
Version two sees improvements in boot times - start up is now five seconds - and the addition of the Clutter interface, which uses a framework to simplify development and provides richer graphics and fluid movements, and dumps the use of widgets. ®