Meet the Distro guy

There's no dearth of Linux distributions. This isn't the first time I've said this. Neither will it be the last. But why the chaos? Why are there more failed distributions than successful ones? Ask the distro guy, Ladislav Bodnar, maintainer of DistroWatch.com. Excuse me if the above sounds like those 15-second commercials during super bowl. As a Linux journalist, DistroWatch is an important tool of my trade. For over half a decade the website has been keeping track of every distribution related activity. And like the many distros it lists, DistroWatch is a one-man show. From its humble beginnings, Bodnar has turned DistroWatch into the most comprehensive, and respected, directory of Linux distributions, it is today.

Below Bodnar talks how he's made a career out of scrolling changelogs and hunting packages in distros.

Read more about this in Mayank's interview with Ladislav Bodnar.

Meet the Distro guy

Excuse me if the above sounds like those 15-second commercials during super bowl. As a Linux journalist, DistroWatch is an important tool of my trade. For over half a decade the website has been keeping track of every distribution related activity. And like the many distros it lists, DistroWatch is a one-man show. From its humble beginnings, Bodnar has turned DistroWatch into the most comprehensive, and respected, directory of Linux distributions, it is today.

Below Bodnar talks how he's made a career out of scrolling changelogs and hunting packages in distros.

Mayank Sharma: We are all aware of how DistroWatch came to be. But I am sure not many recall how the website has evolved over its half a decade period of existence.

Ladislav Bodnar: It started in May 2001 as a single-page table listing a dozen Linux distributions in a table format. As feedback and suggestions started coming in, the page expanded into 13 - the front page with news and the comparison table, plus 12 pages for each of the listed distributions. They were static HTML pages. It wasn't until about a year later when I finally got around to learn some PHP and re-wrote the site to provide dynamic content.

Not much has changed since then. A few cosmetic changes to better organize the available space, a switch to an SQLite database for the news and some other data, and of course the addition of a huge number of new Linux distributions, BSDs and OpenSolaris-based operating systems. In terms of content, the only major new features was the launch of the Distro Watch Weekly newsletter in 2004.

MS: How many distros do you currently list? You hear about new distros launching with much fanfare only to disappear within weeks. Is this something of a new phenomenon?

LB: 530 - of which around 350 are considered active.

No, it isn't a new phenomenon; it's more like a natural evolution of your average tech guy. Perhaps a good example to illustrate this is the now discontinued Beehive Linux, once briefly reaching the DistroWatch.com's top ten. It was a fairly technical distro, mostly for servers, but also with a nice KDE desktop. After a few good releases -- and the rapidly growing amount of time it took to put it all together -- the enjoyment of building it was slowly going away (you could feel it in the changelogs, announcements, etc.). Eventually, in August 2003, the developer wrote a brief post on his web site, concluding it with: "It was fun, but I have better things to do with my life."

This sums up the harsh reality of being a Linux distro developer: the initial excitement over a new idea that leads to a successful release and some fame in the community is quickly replaced by frustration, exhaustion, boredom. Add to it the likelihood of having to spend time supporting the product and getting a few nasty emails from hard-to-please users, and the end result is not hard to predict. It takes a special individual to develop (and keep developing year after year) a successful distribution.

MS: So are you suggesting that distros with multiple developers are less likely to disappear? Is that the reason why some of the most popular distros are those managed and developed by a bunch of people?

LB: Not necessarily. Creating a distro is not unlike learning a foreign language - once the initial enthusiasm is gone, 90% of people just give up. Only if a distro is developed by a strong, motivated individual or by a group lead by a strong, motivated individual, will the distro thrive.

Slackware is essentially a one-man distro (and has been around for longer than any other) and so is the increasingly popular PCLinuxOS.

MS: On the other hand, I think both Debian and Gentoo of-late have run into this "lack-of-motivation" problem, only at a much bigger level and due to different reasons.

LB: I don't think Debian has run into any major problems; in fact, I think people have become more aware of its strength and technologies ever since Ubuntu popularized them. Gentoo is a different story; however, the main problem with Gentoo stems from the kind of personalities it attracts - young, strong, bull-headed individuals who enjoy the bleeding edge and technical nature of the distro. If they can learn to compromise and if they focus on coding and technical excellence instead of whining on public boards, Gentoo can easily regain its former glory.

MS: So how does one select a "safe-choice" distro? Or is that an illusion? :)

LB: On the desktop, I don't think that most technical users care all that much about a "safe choice" and many enjoy "distro hopping", especially those who read DistroWatch. On a server, however, one has to be more careful, but I think there are enough solid options: Debian, FreeBSD, CentOS and Ubuntu are all excellent solutions with long-term support and large user communities.

MS: Every distro on DW gets its own page that lists several features and a snapshot of its package lists. With distros moving on to DVDs, how do you decide which packages to list? How much effort does it take to update/create a new one when a new version or a distro comes out?

LB: I update the package list once a year. I ask the readers of DistroWatch Weekly whether they have any packages they'd want to see listed and tracked. If a particular package gets mentioned more than just once or twice, then it will likely be accepted. Many other readers also email me with suggestions and requests, which is another way of letting me know that a particular packages is in demand.

Updating the distro tables can take anything from a few minutes to an hour. With the traditional distros, such as Fedora or Slackware things are relatively easy since their packages are neatly listed on the CD/DVD image, so with the help of a few scripts the work can be done in minutes. With live CDs, things are different. Many modern live CDs and DVDs use various compression mechanisms, so the only way to extract the package lists is to download them, burn them to a media and boot them. But by far the worst OS in this respect has to be OpenSolaris - I have asked around, but nobody has been able to tell me how to obtain a list of included packages and their versions from an OpenSolaris media or installed system.

MS: Tell us a little about the donations programme. How do you select the projects?

LB: I started the donations programme exactly three years ago and have since donated over US$12,000 to various open source projects. As is the case with the tracked packages, DistroWatch readers either email their suggestions or nominate a project for donation on the DistroWatch Weekly forum. If I don't get any nominations, I simply select a project myself based on my personal preferences.

The only frustrating (and surprising) thing about the donations programme is how little "thank you" one receives after sending the money. Even some big projects, such as Debian, GNOME or Gentoo (which received around $500 each) never bothered to acknowledge the donation in any way. This is frustrating since it feels as if the money never reached the intended recipient and disappeared into someone's back pocket instead. If I knew in advance that a project will not send even a simple "thank you for your donation" email, I would not choose it for a donation.

MS: Is it true that you've tried all the distros that link on DW and then some? Or is it an urban legend? Do you think it wouldn't be such a bad idea if several dozen of them just vanished?

LB: Well, I am sure that there are one or two that I never tried although I wouldn't be able to list them off-hand. Yes, it wouldn't be bad if some disappeared or if people stopped creating new ones for a while.

MS: If you were to categorize distros between 4-5 major categories what would they be?

LB: Historically, I always used to split the distributions into categories according to their package management systems: RPM (Fedora, openSUSE, Mandriva), DEB (Debian, Ubuntu, MEPIS), TGZ (Slackware, VectorLinux, SLAX) and SRC (Gentoo, Source Mage, Lunar). These days, however, the package management no longer plays as important a role as it used to; as the distributions mature, they offer much more painless software management then they did a few years ago.

So the current categorization could be split along the lines of how user-friendly a distribution is. With the ever increasing number of Linux users we are seeing a rapidly growing demand for those distributions that offer out-of-the-box support for three technologies: browser plugins (Flash, Java, video playback), media formats (MP3, WMA, etc.) and device drivers (especially for modern graphics boards and wireless network cards).

In this respect, we can split the distributions based on their target market as:

1. easy-to-use (e.g. PCLinuxOS, SimplyMEPIS, Linux Mint, SabayonLinux, Mandriva's commercial editions and increasingly Ubuntu),
2. slightly technical (e.g. openSUSE, Fedora, Mandriva free edition) and
3. highly technical (e.g. Slackware, Gentoo, FreeBSD, Arch Linux). Then there are specialist distributions, such as those for older computers, live CDs for all kinds of purposes, distributions designed for penetration testing and forensics, embedded distributions, etc.

MS: One of the things that make DW popular is its Page Hit Rankings. Could you explain the dynamics behind it? Don't you find it amazing that many distros have actually tried to "cheat" with their rankings?

LB: Every time somebody visits a distribution page on DistroWatch, his or her IP address is logged. Then once a day a script runs through these logs, counts the number of unique IP addresses, records them in a CSV file, and calculates the daily averages for several time periods: 7 days, 1 month, 3 months, 6 months and 1 year. While these data don't mean all that much, the figures do give an indication what is hot and popular among the DistroWatch readers during a particular time frame and also provides an element of competition among the distributions.

As for cheating, no, I am not really surprised; men are competitive creatures, so attempts to manipulate the ranking can be expected. But we are seeing a change in the way "cheating" is done these days; in the old days, it was usually a crazy distro maintainer or an obsessed fan who wrote a few scripts which repeatedly reloaded certain pages. Nowadays, "cheating" is more subtle and innocent. A few links here and there, a mention in the press release, an email signature with a link to a DistroWatch page, etc.

So it can no longer be classified as cheating; it's more like a gentle effort to add a few points on a daily basis.

But why would some distributions do that? Here is what I was told by a distribution maintainer (who shall remain nameless) who visited a major Linux show and talked to a representative of a very big hardware vendor. When asked about the reason for their choice of Ubuntu as a pre-installed operating system on their computers, the representative of the very big hardware vendor apparently replied: "Because it's number one on DistroWatch".

 

MS: Are you planning to introduce some new features to DW? Also is there any area where you could probably use some help from readers?

LB: I have a very long to-do list of improvements, new features and updates, but unfortunately finding time for all of it is increasingly difficult. So yes, I could definitely use some help. However, over the years I've learned that while there are people who would be happy to help, they usually lack the required skills and that those who have the skills, don't want to work for free. So, realistically, the only way forward is to employ a part-time or full-time coder who would then work on adding the requested new features and implement other improvements. Depending on the income from the site, this option might become a reality possibly within the next year or two.

Share This Article
	Save to del.icio.us