UpFront

Are you still writing programs with the same tools and in the same manner as you did five, seven or even ten years ago? Thinking it might be time to modernize your programming arsenal? Follow along as Collin Park takes a step away from 20+ years of C coding and ventures into Python waters. In his new series on LJ.com, Collin tests out Python by putting it to work solving various puzzles, including the “Coconuts” problem (www.linuxjournal.com/article/8728) and Sudoku (www.linuxjournal.com/article/8729).

Exclusively for LJ.com readers, read an excerpt from Chapter 10, “Adding Your Code to the Kernel”, of The Linux Kernel Primer: A Top-Down Approach for x86 and PowerPC Architectures (www.linuxjournal.com/8730). In our excerpt, authors Claudia Salzberg Rodriguez, Gordon Fischer and Steven Smolski “follow a device driver from how the device is represented in the filesystem...through the specific kernel code that controls it.”

Finally, graduate student Ryan Mauer shares the results of his recent project on Xen. Ryan is a student of our regular LJ.com contributor Professor Richard Sevenich at Eastern Washington University. Ryan intended his project to be “an example of how Xen Virtualization can be utilized to prototype a Linux cluster with only a single physical computer, thereby minimizing the up-front hardware costs, as well as providing a way to perform a feasibility study early in the cluster implementation process.” Read the article and learn the results of his work.

1. Percentage of persons aged 12 to 24 who prefer MP3 players to radio: 85

2. Percentage of persons aged 12 to 24 who listen to satellite radio: 2

3. Thousands of persons in the US who downloaded a podcast during 2004: 820

4. Millions of persons in the US who downloaded a podcast during 2005: 4.8

5. Percentage of podcast listeners in the US who download on a weekly basis: 20

6. Conservative estimate of podcast listeners in the US by 2010, in millions: 45

7. Aggressive estimate of podcast listeners in the US by 2010, in millions: 75

8. Number of Google results for “podcasting” on September 25, 2004: 24

9. Millions of Google results for “podcasting” on December 6, 2005: 78

10. Thousands of Ubuntu/GNOME Linux desktops deployed in schools by Macedonia: 5

11. Number of schools to which the new Macedonian Linux desktops are going: 468

12. Number of computer labs also getting Linux desktops: 182

13. Percentage of surveyed Micro Center customers who are receptive to purchasing a Linux desktop: 75

14. Position of Red Hat in a list of popular Linux distributions: 1

15. Position of Debian among the fastest-growing Linux distributions: 1

16. Red Hat share of Linux Web servers: 34

17. Debian share of Linux Web servers: 25

18. Fedora share of Linux Web servers: 16

19. SUSE share of Linux Web servers: 11

20. Millions of active Debian sites: 1.2

1–7: Bridge Ratings

8, 9: Doc Searls

11, 12: The GNOME Journal

13: Linspire & Micro Centers

14–20: Netcraft

Micro Center is a chain of 19 computer gear stores in 13 states. You find them where techies are concentrated: Santa Clara, Tustin, Madison, Chicago, Denver, Houston, Cambridge and Fairfax. You find them on-line too at microcenter.com.

If you go into their stores, you now find desktop Linux doing something more than bringing down the price of some bottom-end Windows-less PC. Desktop Linux is actually featured. Or, in retail lingo, merchandised.

When the decision to promote desktop Linux was made, Kevin Jones, VP of Merchandising for Micro Center, sent out a letter that said:

This progress isn't limited to the chain's off-line stores. A search for “linux” at microcenter.com brings up 13 pages of results. And, if you want to know one reason why the on-line store performs so well, go do “What's this site running?” at Netcraft.

Eleven servers come up, all running Linux.

Fear, because what is free is never fully appreciated.

—Steve Gillmor, blogs.zdnet.com/Gillmor/?p=189

Remember, Linux is a species, and we aren't fighting anyone here. We are merely evolving around everyone else, until they aren't left standing because the whole ecosystem changed without them realizing it.

—Greg Kroah-Hartman, on Linux-Elitists

I knew Ubuntu (www.ubuntu.org) had reached a milestone when my friend John, a Windows expert of long standing, told me he now tests hardware combinations with Ubuntu. “When I loaded XP on this new box I put together yesterday, it was a nightmare. It couldn't find peripherals, wouldn't load all kinds of stuff. Then I booted it with Ubuntu and it found everything, just like that. So I knew the machine and the peripherals were fine.”

But what's good for experts like John still might not be simple and easy enough for schools. That's why we now have Edubuntu (www.edubuntu.org), a new Ubuntu distro that's customized for classroom use. Says the edubuntu.org Wiki, “As an educator, you'll be able to set up a computer lab or establish an on-line learning environment, in an hour or less—then administer that environment without having to become a fully-fledged Linux geek.”

Edubuntu is built on the 2.6 Linux kernel and GNOME 2.12. It contains more than 16,000 pieces of software and fits on a bootable CD. It is also pointedly noncommercial. The Edubuntu Manifesto begins, “Edubuntu will always be free of charge, and there is no extra fee for the 'enterprise edition'. We make our very best work available to everyone on the same Free terms.” The Manifesto also commits to a regular and predicable six-month release schedule interval. The first official release of Edubuntu 5.10 came out on October 13, 2005. The next release should then be due in April of this year.

A new cryptographic filesystem, eCryptFS, has been sent to the kernel developers for consideration. Phillip Hellewell has been working with Michael Halcrow, who presented a detailed design of eCryptFS at the 2005 Ottowa Linux Symposium. eCryptFS may be “stacked” on top of other filesystems, using the FiST-stackable filesystem framework written by Erez Zadok. The current version supports per-mount decryption only, essentially to support testing and debugging of the basic features. Advanced policy and per-file public key support is planned after some amount of testing can be done on the existing infrastructure. One of the most interesting features of eCryptFS is that it stores its cryptographic metadata within the files themselves, allowing cool features like securely transferring files across untrusted domains.

Mark Gross submitted a special character driver for the Intel NetStructure MPCBL0010 single-board computer that will ship in the fall. The MPCBL0010 is essentially intended for use in telecom devices, which often need to synchronize their operations with other hardware. Mark's driver enables this synchronization via the onboard FPGA (Field-Programmable Gate Array) intended for that purpose. And, a sysfs interface provides easy control over this synchronization to the system developer. Mark also included an ioctl interface, primarily so the driver could be tested under 2.4 kernels, but the ioctl probably will be dropped in any final version of the patch accepted into the 2.6 tree.

The saga of software suspend continues to be broad and sweeping. Rafael J. Wysocki recently posted some patches to split the entire swsusp infrastructure into two independent kernel subsystems. One subsystem is responsible for snapshotting the running system, and the other is responsible for reading and writing the snapshotted data to and from swap. One of the benefits of this division is that it allows a significant portion of swsusp's functioning eventually to be migrated to user-space. Aside from that, the new system allows more memory efficiency, and it does away with certain size limitations and reliance on global variables. Pavel Machek also has gone over these patches and approved them. As we watch big changes like code forks, reunifications, subsystem divisions and so on, it's important to remember that in the beginning, software suspend was deemed impossible because certain hardware state just couldn't be saved. It's no surprise that this feature's problems remain large.

The git revision control system continues to develop at a hot pace. If you haven't tried it, you should. The Ubuntu distribution's kernel development is now done with git, and ethtool development recently migrated to git as well. As a side point, it was pointed out recently to Randal L. Schwartz (now using git for Web site development) that git is binary-safe and easily handles image files, compressed tarballs and anything else one might throw at it. Other folks, like Jeff Garzik, have been experimenting with creating a peer-to-peer storage back end for git, allowing users to query a larger git network, rather than a single standalone system. File renaming in git continues to be a hard sell from Linus Torvalds, as he insists that rename tracking can be done automatically and does not require the user informing the system of a rename. Although no longer the primary git maintainer, Linus continues to influence development strongly, and he remains the biggest evangelist of git features, often pointing out powerful and non-obvious idioms to mailing-list readers.

Adrian Bunk has taken over from Rusty Russell as the Trivial Patch Monkey maintainer. As created by Rusty, the Trivial Patch Monkey collects and submits patches that are so trivial they could not possibly be wrong. Originally, many such trivial patches would get lost in the shuffle. Contributors wouldn't know precisely where to submit them, and the developers who might receive them (including Linus Torvalds) often just let them drop on the floor, choosing instead to focus on more significant work. Rusty decided to pick up these patches and submit them regularly until they finally would be accepted. With each kernel release, he would initiate the semi-automated process of sending these patches along, with the form and timing most likely to appeal to Linus. Now Rusty has apparently moved on, and Adrian will perform a similar task. The Trivial Patch Monkey is not unlike the new w.x.y.z stable tree (dubbed the sucker tree by Linus). Both are largely thankless tasks that lack the thrill of participating in new feature development. And, both make a large difference to the community, though we don't spend much time thinking about them.