The NILFS2 filesystem is a new versioning filesystem, hot off the presses from Ryusuke Konishi. This filesystem does continuous data snapshotting, so that immediate recovery is possible after any inadvertent file deletion or corruption. The project has the support of Andrew Morton, who has put it in his own tree for wider distribution and testing. At this stage, the on-disk data format is nearly stable according to Ryusuke, but that will be a key issue in getting the code accepted into the main tree. Once the code is accepted, any changes in disk format almost certainly will have to include support for all previous formats. In general, this is extremely undesirable in a filesystem, so it's likely that Ryusuke will try to finalize the data format before submitting the code to Linus Torvalds.
Jonathan Corbet has written lots of extremely useful kernel documentation, including O'Reilly's Linux Device Drivers. Recently, he wrote a fairly long intro to kernel development, intended for developers employed by companies who support their kernel work. The goal is to make sure those companies understand what to expect from the developer community and from the relationship between their engineers and that community. This is an excellent document, filled with detailed advice and explanations to help newcomers understand how best to get their features into the kernel. Jonathan has submitted the work for inclusion in the Documentation directory of the kernel sources, though it also may appear on kernel.org at some point.
The old FireWire wiki, having been overrun by spammers, is being replaced. Stefan Richter created ieee1394.wiki.kernel.org, which is already more up to date than the old spammy one, and it's better maintained as well. Those pesky spammers! When we all have nanotech brain implants, will the spammers get into those as well?
Reporting BIOS bugs to user space may be useful or it may just be overkill. Thomas Renninger has been working on this though. Various subsystems, such as ACPI and PCI, can introduce BIOS bugs, and the kernel has to sanity-check all of them. Thomas' argument is that user-space code would get several benefits from having access to the results of this sanity checking. Applications would be easier to test; they'd be better able to respond to the bugs when they were encountered; and users debugging their systems would be better able to identify the problems. Thomas wants to log all of these BIOS bugs to the system log files, where any user-space program could access them. On the flip side, as Bjorn Helgaas points out, the specific log entries for each of these BIOS bugs would have to be maintained individually, and new ones would have to be created by hand. The whole infrastructure would be subject to rapid aging, just like on that Star Trek episode, except without the miraculous cure. But, Andi Kleen thought the benefits would outweigh the risks, offering various implementation suggestions, and even Bjorn had implementation suggestions of his own. So, it does seem likely we'll be seeing BIOS bug logging coming out of the kernel soon.
The MTD subsystem is being worked over pretty well to try to support Flash drives greater than 2GB. Bruce Leonard has been leading this charge, although unfortunately, he hit some technical obstacles when he modified the kernel ABI (Application Binary Interface). This is a real no-no, but as Tim Anderson points out, it may be necessary only to extend the ABI rather than actually change its existing interfaces. Once Bruce and others figure out the right interface for it, it's a dead certainty we'll be getting large Flash drive support in the kernel, probably very soon.
David Woodhouse has been following up on his effort to remove all binary firmware from the kernel. This controversial effort is inspired by the fact that it's weird having lots of binary-only data in an open-source project. But, weird as it is, it's been very convenient for kernel developers to have the firmware in the tree. David's effort involves extracting all those binary blobs into a single firmware git repository, which he has now created. He's also opening up the tree not only to firmware that has been distributed with kernel sources, but also to all firmware everywhere that vendors want to make available to Linux users. So, we'll see what comes of that, and whether they run into similar conflicts as those between kernel developers and GCC developers, where members of each project blame certain problems on the other project, leading to long-term disputes. This is less likely in David's firmware project, as the kernel folks could easily just revert to including their selected firmware blobs in the kernel again if there are any real disputes. Time will tell.
The Dell IdeaStorm site (www.dellideastorm.com) was an inspired move by the company, providing a way for the market to tell a major supplier what to do, rather than the reverse, which has been the default for the whole Industrial Age.
When the site first went up, it sustained what we might call an Insistence on Service Attack by Linux and open-source geeks. Since then, however, the pressure hasn't let up. At the time of this writing (on September 10, 2008), the same kind of demand is there.
What we see with IdeaStorm now is a rolling picture—almost a scroll—of market demand. Here's the current list in the order the items appear on the page:
Put Ubuntu on the list of operating systems when building a PC.
No more plastic wrap, please.
BIOS upgrades that don't require Windows.
Provide Linux drivers for all your hardware.
Standardize power cables for laptops.
Can we get Studio Hybrid with Ubuntu?
There should be an option of having no trialware on all computers.
Please make the Ubuntu XPS Notebook cheaper than the XPS Vista Notebook.
Use magsafe power connectors.
Pre-installed OpenOffice.org | alternative to MS Works & MS Office.
When you choose not to implement an idea, explain why.
Mini 9 Netbook Ubuntu price must be cheaper than the XP price with same config.
Have Firefox pre-installed as default browser.
Switch to LED monitors.
Quit forcing McAfee subscriptions.
Tell us what Wi-Fi chipset a laptop has.
Backlit keyboards.
Stop overcharging on notebook RAM.
To sum up, customers want practical improvements, transparency, promotional crap removal and Linux/Ubuntu support (the latter shows up four times). Maybe some other makers will start listening too.
1. Apache's percentage market share among Web servers in August 2008: 49.82
2. Microsoft's percentage market share among Web servers in August 2008: 34.88
3. Percentage of Netcraft's top ten most reliable hosting companies that run on Linux: 40
4. Position of Linux-based Hurricane Electric among Netcraft's top ten most reliable hosting companies: 1
5. Number of words in the Walt Disney Internet Group's Terms of Service: 5,038
6. Number of words in the AT&T on-line Terms of Service: 10,944
7. Number of words in the Verizon on-line Terms of Service: 8,569
8. Linux percentage share of the smartphone market in Q2 2008: 7.3
9. Soon-to-be-open-sourced Symbian share of the smartphone market in Q2 2008: 57.1
10. Windows Mobile share of the smartphone market in Q2 2008: 13
11. RIM share of the smartphone market in Q2 2008: 17.4
12. Apple iPhone share of the smartphone market in Q2 2008: 2.8
13. Linux adoption percentage at the 14 largest investment firms in 2006: 60
14. Estimated Linux adoption percentage at the 14 largest investment firms in 2008: 72
15. Dollars granted by the National Science Foundation to Polk Community College and University of South Florida Polytechnic to create a four-year Linux computer system administration program: 812,726
16. Size in billions of dollars of the global advertising market in 2007: 600
17. Projected size in billions of dollars of the global advertising market in 2012: 707
18. Size in billions of dollars of the interactive (mostly on-line) advertising market in 2007: 45
19. Projected size in billions of dollars of the interactive advertising market in 2102: 147
20. Annual percentage growth rate of interactive advertising: 23.4
1–4: Netcraft.com
5: disney.go.com/corporate/legal/terms.html
7: www.verizon.net/policies/vzcom
8–12: Gartner, via www.tectonic.co.uk
13, 14: CIO
15: Orlando Business Journal
16–20: The Kelsey Group
WAP Review calls Linux Journal's mobile site, m.linuxjournal.com, “a beauty” and adds, “No tiny dumbed-down mobile site, it features just about everything found on the full site, including archives, commenting, forums and videos in mobile format.”
For that, we also can thank MoFuse (mofuse.com), which hit the streets last summer with a slick Linux-based way of cutting the cruft out of Weblogs and presenting them readably on mobile phones and other handhelds (we gave the company a mention in “Laundering Blog Layouts for Mobile Devices” in the Upfront section of our October 2008 issue).
Now that MoFuse has put some rubber on the road, we thought it was a good time to hit David Berube, Founder & Chief Architect of MoFuse, with a few questions about his fledgling business and Linux's role in it.
DS: How's it going so far?
DB: We are now approaching 14,000 mobile sites, and we just raised a seed round of funding from the Slater Fund in Providence, Rhode Island—our home town. We're moving quickly!
DS: What got you going with Linux?
DB: I chose Linux for a few reasons: 1) it's powerful and can do the job we need at MoFuse really well; 2) it's open, and that is key to how I'm working as an entrepreneur and how our small team can be so efficient; and 3) it's inexpensive—another key factor for a company just starting out.
Linux being free means a lot to an entrepreneur. Using Microsoft would have cost me more to get going when compared to a Fedora install. Also, there are many avenues out there I can reach out to for free and get help—more so on an open platform because it is community-driven. When I have a problem, I can almost always find the answer using a quick Google search. In the past, when I was using Windows servers, the answer or solution wasn't easily accessible.
Basically, Linux has enabled MoFuse to create what it is today and indirectly help foster the mobile Web. I'm sure these are things you've heard numerous times before, but it is very true for MoFuse.
DS: Well, we've heard some of the same rationales before, but fostering the mobile Web from an open-source angle is somewhat new. How are you doing that?
DB: Lots of ways, but one that comes to mind is the OpenX (www.openx.org) ad server. The OpenX team has done a fabulous job with it, and it's a great product. It's not in production, but we are working with it heavily behind the scenes to convert it to a mobile ad server. On our side, we're making modifications so that we can use it as our primary mobile ad server.
DS: How about the larger mobile Web? Any handheld platforms you prefer?
DB: I'm very excited about Android and the prospects for a mainstream open-source mobile operating system using Linux. This will lower barriers for smaller companies and help further the growth of mobile far beyond where it is today—and hopefully, help drive the handset costs down so more consumers can have access to smartphones.
DS: One thing I like about MoFuse is that it cuts the bloat out of blogs and everything else that runs through it. Do you think this will have a positive influence on blog design and send some hints to the authoring software development folks?
DB: I don't know. When I design anything, I do it with two things in mind, the user (viewer) and simplicity. I think MoFuse reflects that, and my personal blog (blog.daveberube.com) certainly does. I like simple; I'm a fan of 37signals because of its take on design, and I try to bring that same take into projects I work on. But, as you know, some of the bigger blogs need to make space for advertising, and that can really clutter a design. There needs to be a balance between design and revenue.
DS: I have a hope that mobile platforms, as they become more popular, will drive development in the direction of simplicity. If that happens, what will you be looking for from other developers, bloggers, Web site designers and your own team?
DB: With anything, you begin to see features creep in. It's a battle every developer wages, and it's not easy. You really need to figure out what is needed and what isn't. The good thing about mobile is that applications and Web sites need to be simple in order for them to function. For example, there is no mouse, so you must make the user interface simple to use in order for users to be able to do the things they need.
DS: Speaking of thin and lightweight, have you checked out Dave Winer's “news river” concept? The best example is nytimesriver.com, which Dave put together for the New York Times, but which it seems to have ignored.
DB: Yes, I have. It's straight to the point, is what it is. Nothing more, nothing less—perfect.
DS: Let's talk about the iPhone. My own view is that Apple has created a very slick data device that also happens to be a phone. It points toward a phone business that needs to be a data business. Meanwhile, the other phone makers have many different devices, with many different SDKs. This encourages developers to come to this one very capable platform, even though Apple is its one huge gatekeeper.
DB: Apple really has created not just a great device, but a whole new concept that flies in the face of the carriers, including its partner AT&T. Jobs has really leveraged the Apple brand to bend carriers to its needs, although Apple did cave in a little with the iPhone 3G and the forced contract with AT&T that subscribers now need to sign.
What I dislike about Apple's iPhone and the App Store are its limitations. Why are there limitations? Why can't a developer create video recording software? Why can't developers create navigational software? Why does Apple get to be the gatekeeper to what gets exposed to the users? Why can't the users be their own gatekeepers?
It also stifles innovation! A company isn't going to invest $100,000 in developing new software for the iPhone, if there is no guarantee that Apple even will approve of the software. Apple wants you to invest your time and money into developing for its platform, but there is always a risk that iPhone users never will get a chance to see it themselves. If that happens, you virtually have no other avenue to get to all those millions of iPhone users, except maybe marketing it to jail-broken phones. I just hope that over time, Apple will open up the platform a little more. We could really see some cool things if it did that.
DS: Meanwhile, on the Linux side, I've discovered some of the phones can't do sound yet. Can they compete? How?
DB: That will change. Sound is moving forward. If you don't have it now, you will have it soon.
DS: And, you think Android will make it happen?
DB: Hope so. The problem Android is going to have is that the carriers are still going to try to lock down the operating system. It's going to be an OS that is fractured, because it's not going on one single device but on many. That means developers are going to have to create applications that work for devices with and without a touchscreen, with and without a QWERTY keyboard, with 3G and Wi-Fi and without. The list goes on. That's a challenge, but hopefully what Android can do is at least create a standard moving forward.
It's up to the carriers. If they don't lock down the operating system and really let it be a true sandbox for developers, I'm positive we will see great innovation. The mobile phone is capable of doing so much. It's a device that we always have with us, and it's always on. It's the most personal computing device we have.
I'd rather not waste my time on nostalgia. I'd rather spend it on hindsight.
—Edward Felten, from his talk at US v. Microsoft, 10 Years Later, at Harvard University
While VMware is in use (www.vmware.com/company/news/releases/cern.html), the primary configuration for machines in the LHC computing grid (lcg.web.cern.ch/LCG) is based on the Scientific Linux distribution running directly on the hardware. This grid is used to receive and distribute the 15PB of data across the 100,000s of CPUs across the world.
—Tim, a commenter writing from a CERN IP address to an InternetNews.com story on the Large Hadron Collider, blog.internetnews.com/skerner/2008/09/large-hadron-collider---powere.html
Our commitment to Linux has not changed....What's changed is that customers will no longer be able to order Lenovo ThinkPads and ThinkCentres with pre-installed Linux via the lenovo.com Web site.
—Ray Gorman, Lenovo spokesman, in an e-mail to Computerworld, computerworld.com/action/article.do?command=viewArticleBasic&taxonomyName=hardware&articleId=9114485&taxonomyId=12&intsrc=kc_top
Apple views tennis-shoe DRM as a way to head off what it sees as a potential plague of sneaker hacking.
—Nicholas G. Carr, www.roughtype.com/archives/2008/09/apple_declares.php
Join us at www.linuxjournal.com/live for our live, streaming video show. Each show, you'll be able to interact with other Linux Journal readers and pose questions to our editors, columnists or special guests. You always can watch the recorded show afterward, but it is far more fun to join in the fun during the live show, so check back on LinuxJournal.com for updates.
We still have our video how-tos and reviews you have come to know at www.linuxjournal.com/video, and we hope you also will join us for our live shows as well. See you there!
Robert L. “Bob” Morgan reminds me why I miss the times when Linux Journal was headquartered in Seattle, not far west of U-Dub: the University of Washington. RL is one of many old UNIX/Linux hands in the Alpha Geek circle there.
I ran into Bob again at Digital ID World in September 2008, in one of the few open-source sessions held there. He sat in the back row, worn laptop on his knees, and asked Knowing Questions. So, I immediately asked him to be the object of our subject for this month. Here's how he ran down his goods:
My laptop is a three-year-old IBM/Lenovo ThinkPad X41 (non-tablet), now running Fedora 9. I have used a series of “ultralight” ThinkPads for perhaps a dozen years (560Z, X21, X31 and now X41), and run Red Hat or Fedora Linux on every one. I do all my work on my laptop and carry it everywhere, so lightness, durability and stability are key. This combination has worked for me.
In my job (identity management architect for the University of Washington and Internet2), I do lots of e-mail (4,000 messages sent per year, average 200 incoming per workday). I use Pine, for much the same reasons as above for my choice of laptop and OS: it's fast, very stable and keeps on working year after year. It's nice to know that the folks who write it work upstairs from me, but I haven't needed any special support.
Maybe the most retro choice on this machine is the Window Maker window manager. It hasn't changed much at all in the last several years, but once again, it's fast and keeps on working. I live happily switching among my 20 virtual desktops with well-practiced keystrokes, and don't miss all those fancy features.
Like anyone else, I use modern full-featured apps like Firefox and OpenOffice.org, but if I'm not reading or composing e-mail in Pine, I'm probably taking notes in Emacs. I guess I'm just a retro kinda guy.
And it's great to have those guys around.