Enterprise Unix Roundup: Debian Thins the Ranks
|Main||In Other News||Recent Updates||Tips of the Trade|
We'll go on the record admitting to a deep admiration for the Debian project and its most famous product: Debian GNU/Linux.
With 8,710 packages and a meticulously coordinated release schedule that encompasses 11 architectures, the project produces one of the best, most stable Linux distributions going, and it does so without compromising a
We'll also on record describing Debian's release schedule as "glacial." The last major release of Debian single jot of the idealism behind Free Software.
One of the most frequently cited reasons for the project's tardiness between major, stable releases is support for eleven architectures: Debian runs on everything from ARM to S/390s. If an official Debian component isn't running on any of those architectures, the distribution on the whole isn't soup.
So many people cheered when members of the Debian release team came back at the beginning of the week with a proposal to scale back a little on what constitutes a "supported architecture" for purposes of a Debian release. The current version of Debian moving toward stable release ("Sarge") would not be affected by the new proposal, but it could have a radical impact on the scope of supported hardware in the next planned release ("Etch").
According to an e-mail from release manager Steve Langasek, the team wants to apply a series of criteria (including whether a particular piece of hardware can still be purchased new, and whether the Debian project itself has the hardware available for binary builds) that could, according to Langasek, "reduce the set of candidate architectures from 11 to approximately 4 (i386, powerpc, ia64, and amd64)."
Everything Debian does is freely available under the provisions of the GNU General Public License (GPL), so even if the project isn't supporting as many architectures as it once did, the raw materials are there for anyone to come along, pick up the pieces (which are well-architected in the generic), and port Debian to their preferred hardware.
The proposal does not, as some reports imply, spell an end to all Debian support for the architectures that don't make the cut: The project would continue to provide some support through its "second-class citizen" architecture, wherein builds for the architectures involved continue to be provided as part of Debian's "unstable" distribution. The unstable branch of Debian tracks the very latest software updates prior to introducing them as part of the testing release. It thus acts as a filter to ensure test branch users are shielded from critical, system-disrupting bugs.
There will be fussing in and out of the Debian project over the proposed changes. The project conducts its business in an open, democratic manner, and it's a stomping ground for some very opinionated people.
The change isn't a small one: Debian has staked much pride on providing a stable distribution for so many architectures, and a few people will be left with less support for their preferred architecture.
But that's where one of free/open source software's advantages comes to the fore: Everything Debian does is freely available under the provisions of the GNU General Public License (GPL), so even if the project isn't supporting as many architectures as it once did, the raw materials are there for anyone to come along, pick up the pieces (which are well-architected in the generic), and port Debian to their preferred hardware. The project would be powerless stop him or her, even if it wanted to.
It's like we pointed out two weeks ago: Linux and much open source and free software are less valuable unto themselves than they are valuable as the starting point for something else. They're often nothing more than freely available raw materials waiting for someone to come along and do something interesting. That's an idea recently echoed (albeit with an eye on money) by Sun's Jonathan "Just Being Quotable" Schwartz, who noted of his company's recently open-sourced Solaris and open source software in general, "the money's not in the access to the product, it's in the services and value delivered around the product."
In the Debian project's case, its release team is proposing to cut down the number of houses it builds with its big pile of raw materials, but it's leaving behind all the blueprints, a respectable pile of wood, and some tools for anyone who wants to come along and build something useful. That's the process Debian has always represented so well, both in its idealism and its product.