UpFront

by Various

UpFront

LJ Index, October 2008

1. Number of Firefox 3.0 downloads in 24 hours starting June 17, 2008: 8,002,530

2. Position of the above download number among Guinness World Records: 1

3. Minimum millions of Firefox users: 180

4. Minimum number of countries in which Firefox is used: 230

5. Millions of dollars in VC funding for open-source vendors in 2007: 328.5

6. Millions of dollars in VC funding for open-source vendors in first half of 2008: 321.3

7. Percentage increase in VC funding in first half of 2008 over 2007: 62

8. Total billions in open-source funding through first half of 2008: 2.80

9. Age in years of a reported bug in BSD discovered in May 2008: 25

10. Linux's market share percentage of counted browser visits in January 2007: 35

11. Linux's market share percentage of counted browser visits in January 2008: 64

12. Linux's market share percentage of counted browser visits in June 2008: 80

13. Linux year-over-year percentage growth in server sales in Q1 2008: 8.4

14. Linux server revenue in billions of dollars for Q1 2008: 1.8

15. Linux-based server percentage of all server revenue: 13.7

16. Number of Linux-based laptops to be distributed to students in the Tamil Nadu state of India: 100,000

17. Price in dollars for Ubuntu Hardy Heron (8.04) in ValuSoft boxes at Best Buy and Amazon: 20

18. Number of Linux distros listed on DistroWatch's Page Hit Ranking: 100

19. Position of Ubuntu among top distros in the six months ending July 16, 2008: 1

20. Position of OpenSUSE among top distros in the six months ending July 16, 2008: 2

1–4: SpreadFirefox.com

5–8: 451 Group

9: TheInquirer

10–12: NetApplications.com

13–15: International Data Corp (IDC)

16: LinuxWatch

17: DesktopLinux.com

18–20: DistroWatch.com

It's All Like...What?

Google phrase searches can produce results that seem like random answers to a Rorschach test—only more amusing. Here are the top results (on a day in July 2008) for “The Internet is like...”:

  • “a vast uncataloged library”

  • “a vagina”

  • “Joey Bishop”

  • “a series of tubes”

  • “Microsoft”

  • “alcohol in some sense”

  • “a newspaper”

  • “going down the Chinese road”

Closer to home, “Linux is like...”:

  • “switching from a car to a motorcycle”

  • “a pizza”

  • “ice cream—too many flavors to choose”

  • “a fixed-wheel bicycle”

  • “turning on your first computer and figuring out what all those weird boxes did”

  • “a common cold”

  • “a whole new species”

  • “Ubuntu, only different”

Our hat tip for this observation goes to reader Mike Warot. (See “What They're Using?” on page 17 for more about Mike.)

Laundering Blog Layouts for Mobile Devices

Whenever I run across something new, cool and Web-based, I check Netcraft's “What's that site using?” to see whether the site's servers are running on Linux. You can't always tell, and the results can be misleading, but it's a good first sniff test.

That's what I did with MoFuse (www.mofuse.com), and it passed. “Linux Apache/2.0.63 (Unix) mod_ssl/2.0.63 OpenSSL/...”, it said.

What's cool about MoFuse is that it launders the layout complexities out of blog posts and turns them into simple lists of linked headlines. It does this free if you like, or you can pay for extra services, such as advertising revenue sharing and URLs that don't have “mofuse.mobi” in them.

MoFuse isn't alone, of course. Mippin (www.mippin.com) has been doing similar stuff for a while, and Netcraft shows them running Linux too (Red Hat, in this case).

And Google, naturally, also has an interface that can turn any blog into “mobile.blogname.whatever”. Of course, that's Linux too.

In any case, the percentage of Web surfing via mobile devices is going to skyrocket—especially after the next generation of unlocked, uncrippled Linux-based devices start hitting the market. Formatting for the hand screen will shift from exception to rule.

They Said It

Our passion is about Linux for human beings, it's not Linux for Linux specialists, or Linux for anything other than the people who we care about.

...our goal, very simply, is to make sure the free software ecosystem can deliver a Mac OS-like experience, or an experience that will compete with the Mac OS. We see Apple as the gold standard of the user experience. We believe that, while it can be a challenge, the innovation inherent in the free software process can deliver an experience that is comparable and in many ways superior.

Proprietary drivers are a horrible kludge; they're a little bit like introducing a cast iron pot into a titanium machine. You have something that is inherently brittle and therefore reduces the value of the whole.

For every artificial scarcity, there's an equal and opposite artificial abundance.

—JP Rangaswami, a talk reported at Reboot10, twitter.com/dweinberger/statuses/844937650

If you provide your customer with solutions, you'll have a well-defined market. But if you provide your customer with the opportunity to create new solutions, your customer will create new markets for you.

—Bob Frankston, private e-mail message

Communications tools don't get socially interesting until they get technologically boring....a tool...has to have been around long enough that most of society is using it. It's when a technology becomes normal, then ubiquitous, and finally so pervasive as to be invisible, that the really profound changes happen....

—Clay Shirky, from his book Here Comes Everybody, page 105

What They're Using: Mike Warot

Linux converts don't always come in groups. Among IT folks, they tend to come one by one. Case in point, Mike Warot, a reader I knew for insightful comments and fun ideas (for example, see “It's All Like...What?” on page 16). Mike's experience is a good sample of what must be happening in countless shops, even though each is different. Now, here's Mike's description of “what he's using”:

I'm a one-man IT staff in what used to be a Windows shop. Today I have two machines on my desk. The one I just let run is Ubuntu. It just runs, 24x7, and it's there when I need it.

The change came once it became apparent that we were going to have to put our Windows servers behind a firewall. I decided to use Linux to make a firewall. I used Red Hat and EBtables to create a transparent bridge to avoid having to re-address our servers and let everything keep its public IP address. This was nice, because it still wasn't necessary to have a firewall, it was just an extra precaution at the time.

That worked well for about a year, until I got worried I wouldn't be able to restart it. That's when a friend told me about IPcop in his work with nonprofits. IPcop is a very easy to configure and maintain Linux-based firewall, which does a great job in almost all circumstances. I used IPcop for a year or so, until problems with multiple Internet connections forced me to seek a better solution.

At the same time I put in the IPcop box, I learned (eventually) two very important facts about Windows 2000 Servers and the Internet. First, if your domain controllers have multiple subnets from which to choose, they will get confused and not be able to find each other (meaning users might not be able to log in). And second, if you followed common sense and used your real domain name for your Windows domain name, you were FUBAR, because active directory then puts all of your local addresses in the DNS, and you can't remove them (or the domain stops working again).

This required another Linux box, for DNS services to the world. It's running Fedora Core 5, but I'm about to replace it, because I don't know how to update the DNS to fix the current exploit. It'll get replaced with a VMware virtual machine running Ubuntu (more about VMware below). I now have an Ubuntu box with iptables (or ipchains, I always forget), which connects us to our three Internet connections.

With the proliferation of servers, VMware's offer of the free VMware Server was a godsend. And, I've since learned that Linux makes a far better host than Windows does. So, I've got some of my VMware servers running on Ubuntu, with Windows and other things inside them.

Ubuntu has made it pretty easy to set up a new machine—not that Red Hat was very hard. But with apt-get, installation of almost any program is trivial. I use an editor called joe to do editing when I don't have the X Window System set up (on some of the older servers), and it's just sudo apt-get install joe, and I'm up and running.

I'm not a guru, but I've learned enough over the years to get around. Linux offers a wider set of tools than Windows ever could. Bottom line: Linux is amazing, and it's getting better all the time.

New Videos on LinuxJournal.com

As some of you may have noticed, there have been a number of new videos on LinuxJournal.com. If you haven't noticed, well, get over to LinuxJournal.com to learn something interesting in one of our new tutorial screencasts. When you are not enjoying Shawn “Gadget Guy” Powers' antics, you'll surely find something new and useful in our new series of videos. Visit www.linuxjournal.com/video to check out the selection. Perhaps you'll learn how to edit video with FFmpeg or something else to add to your bag of tricks.

As always, Shawn will continue to bring you reviews and commentary from his Gadget Guy studio, so keep watching those too. There's so much to enjoy!

diff -u: What's New in Kernel Development

Keeping firmware in the kernel is a somewhat controversial issue, because it means keeping binary blobs of data in an open-source project. But removing it, as David Woodhouse is trying to do, turns out to be controversial also. Folks like David S. Miller are violently opposed to the idea on the grounds that it would break certain drivers (like tg 3) or make them much worse. Also, it takes a part of the kernel that essentially had been approved by Linus Torvalds—even in binary form—and puts it in the hands of someone else. Unlike GCC, libc and other projects on which the kernel depends, this external firmware project would not be anything like a universal tool; it just would be a key part of the kernel that was stored and maintained outside the official source tree.

But, now that the effort actually is underway, it may be unstoppable. A lot of folks find the problem interesting on a technical level, which generally means they'll eventually find solutions that satisfy everyone. For now, it seems as though firmware definitely will be relegated to its own little spot in the kernel sources. From there, it could end up getting a separate git tree altogether.

Willy Tarreau has adjusted the 2.4 release schedule, based on the results of an informal survey of existing 2.4 users. He found that about 80% of current users didn't upgrade, because a lot of their employees or customers would be inconvenienced by any hiccups in the upgrade path or by the downtime required to do the upgrade successfully. The rest of the users could upgrade easily if they wanted to or if they knew how, but they either were running it on their home systems and simply didn't need the benefits of 2.6, or they were running it on firewalls, routers and so on, and the folks maintaining them didn't have a clearly defined upgrade path—although Willy reckoned it wouldn't be too hard to do.

From all this, Willy concluded that he would put out stability fixes more frequently, so people could get what they needed to fix whatever small problems remained on their 2.4 systems, but that he'd slow down on putting out major releases, and do that only for versions with new PCI IDs, big driver updates, compiler support changes and so on. Willy also suggested that someone write up a complete description of the differences between 2.4 and 2.6 and how to address all the problems of upgrading.

It's useful to be able to write to files under CramFS, SquashFS and even mounted CDs, knowing that the changes will not be permanent and will go away after a reboot. Arnd Bergman has recently implemented this kind of temporary write support for CramFS, and Phillip Lougher also is planning to write a similar feature for SquashFS. A lot of folks objected to Arnd's approach though, saying that this kind of thing should be accomplished by using UnionFS to stack something like TmpFS on top of CramFS. But, it turns out that UnionFS is kind of a mess and not really able to do the job well. Some folks, like Phillip, also feel that stackable filesystem support should go right in the VFS, instead of clunking around in a filesystem of its own.

The uproar against writing one-offs for each affected filesystem (CramFS, SquashFS and so forth) subsided when it became clear that CramFS and SquashFS would be the only two filesystems to do this, except for perhaps ISO9660. At least with a small number of filesystems, the amount of duplicated effort would not be too great. But, the people interested in making UnionFS the canonical method of accomplishing this sort of thing apparently have redoubled their efforts. So, it may be that both groups will get what they need soon enough.

Geert Uytterhoeven asked when SquashFS would be accepted into the kernel tree, and Phillip Lougher said that he actually was making some major changes, after which it would presumably take some time for the code to be reviewed by users and any remaining sizable bugs shaken out. Among the changes, Phillip is considering changing the on-disk data layout, which would solve certain problems but break backward compatibility.

The venerable AdvFS filesystem has been released by Hewlett-Packard under the terms of the GPL version 2. This filesystem has been around since the 1980s, supports journaling and file undeletion, and is very fast in general. Nowadays, there are plenty of journaling filesystems, but this release represents a solid and excellent body of code—and a really terrific body of documentation—available to the Open Source world for the first time. The AdvFS release is not, as Linda Knippers from HP has pointed out, a port of the filesystem to Linux. HP sees this release as providing a lot of cool stuff for open-source developers to use in any various projects where they might be useful. But, it's also likely that some kernel folks will want to do a real port. So, we probably can look forward to seeing AdvFS in the main kernel tree at some point.

Load Disqus comments

Firstwave Cloud