The Raw Driver, used to gain direct access to unbuffered I/O on block devices, has been deprecated for a long time, now that the open() system call supports the O_DIRECT option to provide the same functionality. Adrian Bunk has been orchestrating the deprecation and removal process, only to hit a snag at the final moment. As is usually the case with unwanted user-facing kernel features, simply removing them tends to present a dilemma: either users must find an alternative to that feature, or they must no longer upgrade their kernel. Such situations usually result in the feature remaining in the kernel while a grass-roots effort is made to clean up user space. In the current case with the Raw Driver, it turns out that many users still depend upon it, although a lot of them are making efforts to migrate to O_DIRECT as quickly as possible. But with such widespread use, it's also likely the Raw Driver will have to be kept in the kernel for a long time to come.
Adrian also has been continuing his work to remove all OSS sound drivers from the kernel, but it is slow going. There are still approximately 50 OSS drivers to deal with. Some are for hardware that is fully supported by ALSA, and so those can be removed safely. Others have incomplete or broken ALSA equivalents that need to be fixed, and some have no ALSA versions at all. Adrian has been very diligent over a long period of time, tracking down driver authors and bugs, working with users to identify missing ALSA features and making sure that only truly obsolete OSS drivers are removed and not any that actually are still needed.
An old ATI RADEON framebuffer driver, not updated since 2002 and long since obviated by a newer driver, has been patched out of the kernel by Michael Hanselmann. Although the old driver has been marked as old for a long time, the replacement is not perfect either. In particular, David S. Miller has pointed out a bug in the screen blanking routing that can confuse the X Window System under some conditions. But even David favors Michael's patch, as do other big-time kernel hackers like Benjamin Herrenschmidt, so it does seem as though the old driver will be removed before too long. However, Andrew Morton also has said that if possible, he would “prefer to avoid any userland breakage” when removing the older driver.
Jeff Garzik has published the hardware specifications of two previously closed SATA controller chips, Silicon Image's 3114 and 3124 chipsets. Silicon Image graciously gave Jeff permission to publish these docs, presumably after much private discussion. This new documentation also may encourage support for NCQ (Native Command Queuing), used in high-performance data transfer. This kind of openness must be appreciated in a hardware company. It's important to remember that a lot of hardware remains completely undocumented to the free software community, requiring much effort in reverse engineering or else the abandonment of support for a given product entirely.
Although Wim Van Sebroeck has been maintaining the Watchdog drivers for a while now, he has just agreed to add himself to the MAINTAINERS file. Kumar Gala recently asked on the kernel mailing list about tracking down the Watchdog maintainer, and Arnd Bergmann was the one to suggest that Wim add himself to the file.
Kernel configuration is always under scrutiny for ways to simplify and clarify the myriad available options. Recently, Randy Dunlap hit on the idea of migrating SATA configuration out of the SCSI area entirely. SATA does depend on SCSI to provide a function library, but that library could be implemented anywhere, without being tied to SCSI. As Randy reasoned it, there was no reason for users to have to understand this esoteric relationship between Serial ATA and SCSI. And apparently, although Randy himself is not yet interested in seeing this change accepted into the kernel, the idea seems to have general support among kernel developers, and it probably would be accepted if Randy submitted a version that satisfied him.
In early February 2006, the Spain-based company FON (en.fon.com) was three-months old and had just 3,000 “Foneros” when founder Martin Varsavsky announced a $21.7 million investment from Skype, Google and Sequoia Capital.
If all those companies have their way, everybody in a position to use or deploy Wi-Fi Net connections will be a breed of “Fonero”. Varavsky explains, “To us, the world is divided into Linus, Bills and Aliens. A Linus shares his/her bandwidth for free with other Foneros, Bills share their bandwidth for a small fee, and Aliens don't share their bandwidth at all.” Because Aliens are those creatures called customers.
And those needn't be just geeks like the readers of Linux Journal. Ethan Zuckerman, in his blog My Heart's in Accra (www.ethanzuckerman.com/blog/?p=363), explains why he's both a Fonero and on the company's advisory board:
There's a philosophical bias to many of these projects—a belief that Internet access is an inalienable right and should be free—that I find charming, but totally impractical for the parts of the world I'm most concerned about.
In Africa, bandwidth isn't cheap. Entire universities run on less bandwidth than I have coming into my house on a DSL line. Being altruistic and leaving your wireless access point open in Africa is pretty much a guarantee that you're going to end up with other users abusing the limited bandwidth you have. It's important that African users have the opportunity to share their bandwidth in a way that allows for “bandwidth shaping”—sharing some bandwidth with other users and retaining the rest for your own needs—and billing, so other users can share the cost with you. FON's current software isn't optimized for this situation yet, but it's close, and FON is engaged with the issues in a serious and sustained way. I predict that FON is something I'll be able to pitch enthusiastically to African friends in the very near future.
To run FON, download software based on Sebastian Gostchall's DD-WRT open-source project (www.dd-wrt.org). And, you run it on a FON-compatible router. Right now, that's a Linksys WRT54G/GS/GL (versions 1x to 4x), which are the ones with Linux inside. The first 3,000 are being sold far below cost. Those may be gone by the time you read this, but the company is sure to make it as easy as possible to become a Linus, if not a Bill.
Will Linux reach mainstream desktops and laptops without a major vendor making the push? Several vendors have recently stepped up to answer that question.
At CES in January 2006, Google cofounder Larry Page made a public show of his company's support (sans details) for MIT's $100 laptop, designed to “revolutionize how we educate the world's children”.
At the end of January, Red Hat announced support for the project as well. At the time of this writing, the company is working on adapting Fedora and plans to make the project an open and public one. The company also signed on as a platinum supporter of the Desktop Linux Summit, an event Linspire launched three years ago and still runs.
The New York Times also reported that Nicolas Negroponte, who is running the $100 laptop project, is close to lining up $700 million from seven countries—China, India, Brazil, Argentina, Thailand, Egypt and Nigeria—interested in buying 7 million of the units. A Taiwanese manufacturer was also reportedly lined up.
Meanwhile, Nat Friedman showed off Novell's Linux Desktop 10 in Paris. He played videos and MP3 music files (with Banshee, Novell's own player, using licensed patents), downloaded pictures from a digital camera and exchanged photos with an iPod. He also showed off XGL, an open-source graphics subsystem. Right now, it's on track to be available by the time you read this.
And, of course, the noncommercial open-source projects—GNOME, KDE, freedesktop.org (freedesktop.org) and so on—continue to move forward.
1. Smallest number of Weblogs that mention “open source” per day: 500
2. Largest number of Weblogs that mention “open source” per day: 1,050
3. Smallest number of Weblogs that mention “linux” per day: 1,250
4. Largest number of Weblogs that mention “linux” per day: 2,600
5. Percentage of smartphones shipped with Linux in Q1 2004: 3.4
6. Percentage of smartphones shipped with Linux in Q1 2005: 13.7
7. Percentage increase of smartphones shipped with Linux between Q1s 2004 and 2005: 412
8. Current Linux percentage share of advanced mobile OSes: 17
9. Projected Linux percentage share of advanced mobile OSes by 2009: 29
10. Billions of mobile phone subscribers worldwide by late 2005: 2
11. Number of top 500 supercomputers that run on Linux: 360
12. Percentage of top 500 supercomputers that run on Linux: 72
13. Number of top 500 supercomputers that run on Linux distros: 30
14. Percentage of top 500 supercomputers that run on Linux distros: 6
15. Total number of Linux-based supercomputers in the top 500: 390
16. Percentage of Linux-based supercomputers in the top 500: 78
17. Number of Linux-based supercomputers in the top 10: 5
18. Position of the CNK/Linux-based IBM BlueGene/L in the top 500: 1
19. Growth rate in size of the CNK/Linux-based IBM BlueGene/L in the last year: 2
20. Top Linpack performance of the BlueGene/L, in teraflops: 280.6
1–4: Technorati (during the month of January 2006; numbers rounded to nearest 50)
5–7: Gartner, via Linux Devices
8, 9: TDG, via Linux Devices
11–20: Top 500 Supercomputer Sites (t500.org)
Krugle is a new search engine just for source code and other technical stuff. Ken Krugler, company founder and CTO, puts the appeal in simple terms, “Krugle is a search engine for programmers.”
I was at a conference in San José when Krugle CEO Steve Larsen showed the beta version of Krugle to Bill Weinberg, an old friend who now works as an Open Software Architecture Specialist at OSDL. “I have to have this”, Bill said. Then, when Steve Larsen continued with the demo, Bill added, “No, you don't understand. I need this.”
See if it hits you the same way. Check it out at krugle.com.
For the past six months or so, Ken Hess has been conducting an on-line Linux Consultants Survey to gather consultants' opinions on Linux, both its current state and its future. Now, he's sharing the results of that survey with linuxjournal.com readers (www.linuxjournal.com/article/8873). Based on their customers' experiences, find out what Linux pros are saying about Linux in the data center, as a server and on the desktop.
It seems as though every branch of government spends countless hours and money on its voting system—collecting ballots, counting ballots, recounting, recounting and recounting—and we the people still can't trust the results. Clearly, closed and proprietary systems aren't working, so why not extend democracy to the voting system itself and make it open source? In “The Politics of Honest Voting” (www.linuxjournal.com/article/8872), LJ Publisher Phil Hughes outlines what an open-source voting system might look like. Share your thoughts on the matter, and get involved with turning the current system on its head.
Senior Editor Doc Searls is blogging now on linuxjournal.com, bringing breaking news and commentary on Linux business, trends and evolution. Bookmark this page www.linuxjournal.com/blog/800285 to go straight to his blog, or sign up for the linuxjournal.com RSS feed at www.linuxjournal.com/xstatic/aboutus/rss_page to be notified when a new entry is posted.
Great things are not done by impulse, but by a series of small things brought together.
—George Eliot or Vincent van Gogh, on Google (many quote sites on the Web are split between the two)
All large systems that work start as small systems that work.
—Stowe Boyd
UNIX is basically a simple operating system, but you have to be a genius to understand the simplicity.
—Dennis M. Ritchie, www.brainyquote.com/quotes/authors/d/dennis_ritchie.html
Software wants to become worthless without skilled attention.
—Don Marti (in a conversation)
Skill without imagination is craftsmanship and gives us many useful objects such as wickerwork picnic baskets. Imagination without skill gives us modern art.
—Tom Stoppard, jon.linuxworld.com
When you ask a question about an open-source product, ask the community, not one specific person. When you ask for one person to answer the question, then other people who may know the answer, might not help (in fact they almost never will, assuming you had some reason to want to know the answer from this one specific person). I've been doing this for many years. People almost never want to hear this, so I usually just ignore the questions, even if they have easy answers, because I want a community to develop, one where people help each other. That's the only way it can grow. And I want that kind of growth even more than I want you to get over this particular hurdle.
On the other hand, if you see a newbie ask a question of someone specific, and you know the answer, and you are not the person he or she asked, go ahead and answer it. Assume the person just wants the answer, not really from anyone in particular. If they complain that your name isn't Linus or Brian or Alice, you can tell them that's true, but the answer is still the right one.
—Dave Winer, www.scripting.com/2006/01/14.html#When:8:18:54PM
