LWN.net Weekly Edition for June 1, 2006

Ubuntu Dapper and the distribution business

Ubuntu's "Dapper Drake" release - more prosaically known as "6.06 LTS" - is due on June 1, and may well be available by the time you read this article. A distribution release is not a particularly rare occurrence in the Linux community, but there are a couple of things about Dapper which are just a little bit unusual and worthy of note.

The "LTS" in this release's name stands for "long term support"; this distribution comes with a promise of security updates for five years (on server systems) or three years (on desktop systems). Exactly how that distinction will be made is not entirely clear; one assumes that, for example, graphical mail clients will go unsupported in June, 2009, while mail transfer agents will continue to get updates into 2011. That is the longest credible support promise ever made for a free distribution, and it may change the commercial landscape in interesting ways.

There are many situations where the deployment of a Linux system makes a great deal of sense. In many of those, one wishes to start with reasonably current software, but to not have to worry much about upgrades for a long time thereafter. Web servers, print servers, database servers, kiosks, point of sale systems, and more all fall into this category. Once the system works, any sort of software change offers downtime and the risk of problems, but little in the way of advantages - except, of course, for security fixes. Anybody planning such a deployment must consider how the system will be supported and kept secure through its operating life. In recent years, the available choices have fallen into these categories:

• An entirely free distribution (Fedora, Debian, OpenSUSE, etc.) can be used. The price is right, and the quality of the software tends to be high. The support window for these distributions tends to be short, and, for some of them, unpredictable. Keeping a Fedora Core system secure can involve upgrades twice a year - not an appealing option for a system which is supposed to be stable and "just work."

• The "Enterprise" offerings from Red Hat and Novell come with long support promises; there are, undoubtedly, still plenty of systems running 2.4.9 kernels on RHEL 2 with uninterrupted support. These services can be expensive, however. For many customers, a support subscription is easily justified and worth every penny. But others will find that cost hard to swallow.

  Some try to get the best of both worlds through enterprise clone distributions like CentOS. By all accounts, the CentOS team has done a top-quality job with its distribution, but anybody contemplating a long-term deployment will have to be convinced of the project's long-term future and be able to overcome qualms (if any) about free-riding on the enterprise distributions.

• Security support can be managed in-house. This approach requires a significant investment of time by a skilled administrator or developer, however, and is thus far from being free.

Ubuntu's five-year guarantee provides another choice: install Dapper, and obtain updates until 2011 with no costs at all. The existence of the Ubuntu Foundation, with its $10 million nest egg, helps to make that five-year promise credible, and Ubuntu's record with security updates has been, so far, quite good. So it would not be surprising to see significant uptake on Ubuntu's promise. Whether those new Ubuntu users will come at the cost of the enterprise distributions, or whether they are mostly people getting away from the (relative) upgrade treadmill of the free distributions, remains to be seen.

That leads to the other interesting aspect of this release: the increasing friendliness between Ubuntu/Canonical and Sun Microsystems. The two have just announced that the Dapper release will include a version for Sun's new Niagara SPARC architecture, and Sun executives are issuing quotes on how important a distribution Ubuntu is. Clearly something is going on here.

Sun's troubles in recent years have been well documented; to a great extent, Sun's customers have been steadily turning into customers of the enterprise distributions. To Sun, Ubuntu may well look like an opportunity to poke holes in the revenue streams of its main competitors. Ubuntu, in turn, may see Sun's support (and the Niagara port) as a way to gain a foothold in the server market. If Sun's new servers find customers, Ubuntu will be the obvious distribution for any of those customers who wish to run Linux.

How all of this plays out will be interesting to watch. Ubuntu's past releases have certainly been popular; if Dapper holds together well enough (and the initial signs are good), it may be the best-received Ubuntu release yet. If so, Ubuntu may well change the shape of the Linux distribution landscape.

(For those who are interested in what's actually in the 6.06 LTS release, the "testing Dapper" page has a lot of information and screenshots).

Comments (21 posted)

The end of the JPEG patent - sort of

Forgent Networks is a company which would easily qualify as a patent troll for many observers. This small company picked up a data compression patent in 1997, and has been busily using that patent to shake down corporations ever since. Since this patent is said to cover the JPEG image format, there is a wide list of possible victims to choose from. Those victims have dropped more that $100 million into Forgent's bank account, and Forgent currently has litigation outstanding with some 30 companies.

The Public Patent Foundation chose this patent as one which was vulnerable to a challenge. The Foundation's work bore fruit on May 25, when the US Patent Office issued a ruling on the Forgent patent [PDF]. The resulting press release from the Public Patent Foundation was triumphant:

It is worth noting that Forgent had a different spin on the ruling:

Anybody wondering if the world is now safe for JPEG users will clearly need to look beyond the press releases and dig into the patent and the USPTO ruling directly. The short story is that, while the independent claims of U.S. Patent 4,698,672 have been invalidated, many of the more-specific dependent claims remain standing. Consider, for example, claim 1:

What the Public Patent Foundation asserted is that this claim - covering a fairly basic run-length encoding scheme - had already been claimed by another patent: #4,541,012 by Andrew Tescher. The Patent Office agreed, and ruled that claim 1 was invalid.

The story does not stop there, however. There are a number of dependent claims which make claim 1 more specific; these include:

Claim 3 (adding a sign value) was also rejected, but claims 2, 4, and 5 were upheld by the Patent Office. The same pattern persists through the remaining claims: the independent claims were rejected, but the more-specific versions were allowed. That is why Forgent proclaims that the majority of its claims had been upheld.

So, to a great extent, the Forgent patent survives, having lost only the most general of its claims. We asked Dan Ravicher of the Public Patent Foundation whether this ruling was enough to remove the threat against JPEG users; his response was:

Whether the remaining claims in the patent are applicable to the JPEG standard is a matter for the courts to determine - and, given the thirty-some outstanding cases, the courts will certainly have the opportunity to do so.

There is one interesting additional factor which, thanks to the Public Patent Foundation's work, may just come into play here. Forgent's patent was originally filed from a company called Compression Labs, Inc. It turns out that the Tescher patent, which provided the prior art used against Forgent's patent, was also developed at Compression Labs. In other words, when Compression Labs filed for the patent now being wielded by Forgent, it must have known about the existence of the prior art, since it had patented that prior art itself. But Compression Labs did not disclose that prior art to the Patent Office. Failure to disclose known prior art is a violation of the Patent Office rules. It seems likely that defendants in Forgent's litigation will find a way to let their respective courts know that the patent at issue was obtained in bad faith.

Comments (4 posted)