UpFront

For more than a year, Terence Ripperda has been thinking about adding Page Attribute Table (PAT) support to Linux. Back in mid-2003, the issue did not seem so important to him, because only the AGP aperture and framebuffer really would benefit. With PCI Express systems coming out, however, the old workarounds are too slow. PCI Express lacks AGP's central aperture that can be marked WC (write-combined) in certain cases. Instead, individual memory pages must be marked as WC separately in the page tables. This cannot be done efficiently using the old ways, but by adding PAT support, a high level of efficiency still can be attained. Terence has been working closely with Andi Kleen to get something ready for inclusion in the kernel. Other folks, such as Andy Whitcroft, also have been looking into adding PAT support; so, one way or another, it seems that this enhancement will go through.

Maneesh Soni, acting on a tip from Alexander Viro, has taken on an annoying sysfs bug. Apparently, symbolic links in sysfs have been following the target that existed at the time of the link's creation no matter what, even when the intended target changes during use. A patch posted by Maneesh causes the target to be recalculated every time the link is read. Maintaining a consistent and proper sysfs interface is crucial to providing a clean global interface into the kernel. The procfs mishmash is one of the main reasons sysfs was created in the first place.

Carl-Daniel Hailfinger secretly has been implementing a successor for ATA RAID in 2.6, called raiddetect, and finally unveiled his work for comment in April 2004. Raiddetect is intended to identify vendor software RAID superblocks, verify their validity, group them by vendor and set them up for use. Although Carl-Daniel's work admittedly could have been done as an extension of Wilfried Weissmann's EVMS plugin, Carl-Daniel opted for the easier method of starting from scratch—though he affirms that his work could be incorporated into EVMS even now if folks preferred that. So far, there doesn't seem to be much dissent, and people like Jeff Garzik are filled with praise for Carl-Daniel's work.

Marcelo Tosatti has decided to merge Serial ATA (SATA) support into the 2.4 tree—probably one of the last big features to go into 2.4. Marcelo says the reason for this is that a lot of new computer systems are shipping with SATA-only disks, and until 2.6 becomes fully stable it makes the best sense for 2.4 to be able to run on these new systems. The decision met with some criticism, most notably the counter-contention that the 2.6 tree is plenty stable and specifically includes the same security fixes that recently went into 2.4. Regardless, it seems that there is at least some good reason for including SATA support in 2.4, and Marcelo apparently fully intends to do so.

Herbert Xu has written some code to allow Software Suspend to be compiled as a module, but with all the hubbub over whether loadable modules should be unloadable at all, the question of whether a given feature should be compilable as a module is becoming less critical. Some developers, including Alan Cox, have been leaving modularity support out of their driver work, preferring to get the basic behavior correct and worry about less important modularity features later. Software Suspend is apparently one of these as well, and folks like Pavel Machek and Nigel Cunningham have expressed their doubts that a modular Software Suspend is more valuable than the nonmodular version. The whole question of modularity seems up in the air in the 2.6 series, because the ability to load and unload modules always has been dear to the hearts of many users, and its disappearance has been more for the convenience of developers than for users. If a clear solution to the problem of how to unload modules properly is found, it is likely that 2.7 would see the reappearance of that feature.

From April 29 through May 2, 2004, the Center for Arts and Media Technology (ZKM) in Karlsruhe, Germany, hosted the second International Linux Audio Conference. Developers from around the world presented open-source software for hard-disk recording, software sound synthesis, music typesetting, digital audio signal processing and many other sound and music-related areas. Several concerts and a sound installation demonstrated how this software can be used for composition and production. Linux-based music and sound production hardware, the Lionstracs Mediastation and the Hartmann Neuron, also made an appearance. Visitors were able to set up their own computer-based Linux audio systems with the assistance of experts from the AGNULA/Demudi and Planet CCRMA distributions.

The conference showed considerable growth over last year's successful event. More than 30 topic presentations took place, and many extra sessions were held to accommodate requests for more information about particular topics. All presentations were given in English and were comprehensible and engaging, a testament to the linguistic abilities of the presenters, many of whom were not native English speakers. Highlight topics included Paul Davis' demonstration of recent advances in his Ardour digital audio workstation project, Steve Harris' exposition of the JAMin audio mastering suite, Fons Adriaensen's demo of his Aeolus pipe-organ emulator and Stefan Kersten's introduction to his work with the SuperCollider3 sound synthesis environment.

Other memorable presentations included Orm Finnendahl's explanation of his use of common UNIX tools such as sed and awk in his compositions, Ivica Ico Bukvic's thoughts on getting Linux into school music curricula and Dave Topper's report on his GAIA, a graphic front end for sound synthesis languages.

With four concerts, a dance, ongoing hardware demonstrations and various workshops demonstrating the use of Linux audio software, there was ample opportunity to hear how Linux audio software has evolved. Huge kudos to Frank Neumann, Matthias Nagorni and Goetz Dipper for keeping everything running smoothly. Vast thanks also to SuSE and ZKM for their support. See linux-sound.org for links to software, recordings and a conference program. Next year's conference already is scheduled for ZKM again.

I always know when I get an e-mail from Murray Gray, because he writes to me from the future. He's in Australia, near the leading edge of tomorrow. I'm in California, which is yesterday, relatively speaking. Murray's subject, however is instantaneous. He's into IM (instant messaging). And, here's what he's doing about it, with a new project called Qunu (www.qunu.com):

Qunu traces back to an idea by Joseph E. Trent on the (now defunct) BeNews forum in January 2000. The idea piqued the interest of Helmar Rudolph (formerly of Opera Software and Sonork) and Murray, who joint-financed development by Justin Kirby of openaether.org using his in-house toolkit for XMPP/Jabber development.

Qunu is based on Jabber's XMPP (eXtensible Message and Presence Protocol) and Mozilla. Here's Murray's technical case:

Murray invites readers to download the code, play with it and contribute improvements back to the new community.

If you've been reading about the 2.6 kernel in Linux Journal but waiting until a distribution integrates it nicely before you take the plunge, have a look at SuSE 9.1 Professional.

The install is nothing special, which is a good thing considering the current standard for high-end Linux distributions. A great feature is the check for any updated packages from the Net before letting you out on your own. If a security problem comes up, new SuSE 9.1 installs from CD will get the fix as part of the normal install.

KDE 3.2, the default desktop environment, is slick, with features including easy CD burning, a file manager that lets you drag and drop files anywhere you're allowed to sftp to and great GNU Privacy Guard integration in KMail.

The hard part about doing a Linux distribution, however, is hardware support, and this is where SuSE 9.1 really stands out. SaX, SuSE's X configuration tool, set up a multihead display and a Wacom tablet with point-and-click ease—no configuration or documentation reading required. USB printers and storage devices on our test system also simply work.

Although the price is higher than a lot of distributions, the box set includes good printed user and administrator guides, along with install support. We recommend this distribution for power users of other OSes who are trying Linux for the first time.

SuSE has a good record for laptop support, too, so watch for more on SuSE 9.1 from Doc Searls, who is running this distribution on his IBM ThinkPad.

Just as in the 1990s the COTS sector caught up with the military sector in applications of cryptography, this decade will see the self-same overtaking but this time of traffic analysis. You do not need to examine content if you can deploy enough sensors and make sense of their findings.

—Dan Geer, on the Politech mailing list

Brazil hosted the recent Fourth International Forum on Free Software, held in the World Social Forum's stronghold of Porto Alegre. There, jazzy pop star Gilberto Gil, now Brazil's minister of culture, promised to “tropicalize digitalization”, presumably a reference to bridging the developed and developing worlds. If Extremadura is the harbinger, those tropicalized digits will be globally connected, fiercely patriotic, and free as sunshine.

—Bruce Sterling in Wired

Many on Jack Valenti's side of the divide treasure their creative freedom and fight like dogs against any who would block it. They would never dream of permitting a system in which every film had to be approved by the state, but they are advocating a system in which every program has to be approved by the state, because a lot of them think that all programs come either from faceless corporations like Microsoft or from criminal vandals.

We software creators need to insist that creative applies to us.

—Joe Buck, in a comment on lessig.org

Nat [Friedman] used to say “If you write a thousand lines of code, you are violating someone's patent today.”

The picture is not pretty for anyone in the software industry.

But this is similar to what happens to biology students: in their first four semesters as they learn about all the dangers, infections, vectors for infections and bacteria, they stop eating everything, they start washing their hands with special products, they double clean their utensils, they wash their fruits ten times a day.

Two years later they are eating food with their bare hands again.

—Miguel de Icaza, primates.ximian.com/~miguel