Moblin 2.0 Linux goes alpha (again)
Novell jumps aboard the bandwagon
Novell is swearing its oath of fealty to the Moblin variant of Linux for mobile computing devices, based on Intel's Atom low-powered processors and, soon with the Moblin 2.0 release, netbooks.
Intel launched the Moblin project back in July 2007 and got Moblin 1.0 into the field in April 2008, concurrently with the launch of the Atom processors. These are cut-down variants of the old Pentium-style processors that consume very little electrical power (4 or 8 watts in the current 1.6 GHz single-core and dual-core chips) and yet provide enough computing power to be useful in many kinds of devices.
The original Moblin project was based on the Ubuntu Linux kernel, itself a variant of Debian Linux. Last July, Intel tweaked the project to include a kernel derived from Red Hat's Fedora development Linux and a subset of its Gnome application stack. The pre-alpha Moblin 2.0 release came out in January. (This wasn't pitched as being pre-alpha or even alpha code, but this is apparently the case.)
Despite its Ubuntu and Fedora heritage, Novell seems to be thrilled to be backing Moblin 2.0, which Guy Lunardi, director of client preloads at the commercial Linux distributor, says will be announced as an alpha today and which is expected to begin a normal beta testing schedule in a few weeks.
The exact scheduling of the beta program is a work in progress right now, because the Linux Foundation - the place where Linus Torvalds, the creator of Linux, gets his paycheque - took over the steerage of the Moblin project from Intel earlier this month.
The chip maker has figured out that it is difficult to get original equipment manufacturers (OEMs) and original design manufacturers (ODMs) who design, build, and resell all manner of computing devices to back a project when it is vendor controlled. And despite Intel's goal to have Moblin promote its Atom low-powered processor, it seems extremely likely - one might say inevitable - that if Moblin is to survive as a viable project for embedded and mobile computing devices, then it will probably have to support ARM and PowerPC chips, too. And maybe MIPS chips as well. But thus far, neither Novell nor Intel have talked about that prospect. But that is what independence from Intel means.
This may come out in a few weeks when the Moblin project releases its first product roadmaps in the wake of Linux Foundation taking it over. For the moment, Moblin is using an internal build system created by Intel and is doing nightly builds on Moblin 2.0, but according to Lunardi, this software does not have a desktop interface or its application stack finalized as yet.
Novell is now contributing code to the Moblin project, adding in code for windowing, email and media management from its openSUSE development project, and said today that it will create a Moblin-based "product" for netbooks that it will peddle to OEMs and ODMs. This product will not, according to Lunardi, carry an Intel or Novell brand, but will instead allow these OEMs and ODMs to weave their own brands into the code and basically insulate the user from even knowing they are on any particular operating system.
Novell is also setting up something called Open Labs in Taiwan, working in conjunction with Intel and the Taiwanese Institute for Information Industry, which have already set up the Taiwan Moblin Enabling Centre, to validate Atom/Moblin products.
Taiwan is key for two reasons. One, a large percentage of cell phones, handhelds, nettops, netbooks, and laptops are manufactured in the country. This means there is a huge community of OEMs and ODMs that are looking at chips and operating system platforms for computers that don't fit the usual desktop and laptop PC definitions. Moreover, back in early March, Taiwan Semiconductor Manufacturing Corp inked a deal with Intel that allows it to create customized versions of the Atom chip.
With a truly free Moblin operating system and customizable Atom processors, there is a better chance that OEMs and ODMs as well as other providers of embedded computing devices will take a harder look at Moblin.
It is harder to see what Novell gets out of Moblin until you realize that OEMs and ODMs don't want to manage the updating and patching of Moblin or the optimization of the operating system for particular kinds of devices. So Novell is pitching its online Yast Update tools as part of the Moblin product it will put together. But don't expect to see a SUSE Linux Enterprise Mobile distro. Novell seems content to work behind the scenes for now. ®
why not DSL and save another fork
One more small footprint distro to ignore.
Already plenty exist. With a linux having a dynamic linking kernel, why do special branches on something largely commodity, like Intel chips.
Moblin is fine.
..to see this get somewhere, but no doubt MS has a plan for killing this.
It's definitely going somewhere. One of the major reasons why the numbers of Linux being installed on Netbooks didn't grow with the number of Netbooks being sold is because each Netbook variant also produced their own custom operating system.
Sure the hacked up versions of the Linux desktop that made it's way into various Netbooks were nice to look at and use, but to be really useful users are still going to want to install their own applications and games and whatnot. There is only so much many people you can keep happy with old versions of Firefox and OpenOffice.org.
Advanced users, much to their dismay, found that installing standard Linux distributions required massive amounts of tweaking to properly support their Netbook's hardware.
Look at this way:
You see one of the biggest problems with Linux, in terms of hardware support, isn't that hardware support sucks... it's just that hardware support lags so heavily.
Think about it. The 'proper' way to get new support for hardware in Linux is to go through a process were you submit your patches to the Linux kernel folks on their mailing list. Once you get their attention then you will have to edit your code and resubmit it 3 or 4 more times before it's close enough to a good shape to were they will take it in and introduce it.
And typically, because of trade secrets, support for new hardware doesn't happen until AFTER the hardware is released. This is much less bad then it used to be (once people figured out that Linux developers had no problem with singing NDAs).
But lets assume...
So Asus releases a new Netbook. It requires modifications to the Linux ACPI stuff to support the new variation. It also needs a new driver for the ethernet and a update to the driver for the wireless and webcams.
Alright. In Linux that is easy to do. Hackers have all that stuff working on patched kernels within weeks of the new hardware release.
So after a couple month stay on the in a development branch of the kernel the new code is given the go-ahead to be entered into the mainline branch of the Linux kernel. That takes about 2 months.
However the kernel doesn't do releases every month. Each new kernel version is only released every 3 months, or so.
So that is 2 weeks + 2 months + 3 months release cycle. So we'll assume that so far about 4 months have passed since the new Netbook has released.
But that means that only benefits users that build their own kernels.
For most users they use pre-compiled kernels from their distributions. Ubuntu and Fedora only release new systems every 6 to 8 months.
SOOOOO... Between getting the netbook released and getting new hardware support into distributions means that your looking at about a 4 month to 10 month lag here.
What this gets you is that:
A. Users can't install new applications on their hacked up Netbook-specific distribution easily.
B. To install new applications they will have to install a different Linux OS. Like Ubuntu... however Ubuntu wont' have hardware support built-in for another 8 months, at least.
What makes it worse is that most people won't know this. They'll try to install their own Linux version if they can get that far (which most won't) and then they'll see about 15 pages of Wiki horseshit that they will have to wade through to get good hardware support again.
Sooooo... this means 'Fuck Linux I am using XP'. People need to get work done.
With Moblin they will provide all the hardware support for Netbooks out-of-the-box.
They also will provide a build service that ODMs and OEMs can use to build their own 'custom' version of Linux. As a added bonus between making the OEM's life easier and Linux much cheaper to develop the build service will make sure that all the APIs are there and unified for the Moblin platform.
This means that you will have a standardized set of APIs for the Linux netbook folks that will enable users to easily install new applications, which will have top-notch support for all their hardware.
Soo.. Linux + Top notch hardware support + Easy to install Applications and standard API for ISVs to target.
This is exactly the advantages that Windows XP has over Linux on the netbook desktop....
And even if Moblin dies a flaming death it's improvements to boot performance, drivers, and other things (like GTK CLutter) are being adopted by other distributions as fast as they are being introduced by Moblin.
Dazzelled & Confused
As usual with shinnnnny! new tech I'm dazzelled and confused. Yesterday /. ran a piece on a heretical fork of glibc. See: Switching From Glibc To Eglibc
"Aurelien Jarno has just uploaded a fork of glibc called eglibc, which is targeted at embedded systems and is source- and binary-compatible with glibc." http://developers.slashdot.org/article.pl?sid=09/05/06/2050216
I'm now married into the Atom processor family and lov'n it. Furthermore Debian/Ubuntu is my Linux flavour mix of choice, but I'm wondering if it's too early to look at jumping onto any one bandwagon when it comes to an OS running on top of my darling little chips. On my Acer Aspire One I've stuck with the fedora6 variant the netbook came with, but I'm gearing up to build a few N330 boxes and would like to start off with at least a murky idea as to the best Linux distro to run. Has the reg run any articles on the whole sordid affair that I may have inadvertently missed? On /. venerable old timers with 3 digit UID # were throwing up their virtual hands in disgust and saying the whole kernel had to be shut down and the kernel and lib redone and properly documented (no proper Linux Luser has ever RTFM so I don't see that proper documentation will do any good).
Any ideas? Guidance? Grace from above?
PH because this little mess has me feeling disturbingly empathetic toward her