UpFront

They are struggling with not so much open source, per se, but rather they are no longer the low price solution. In the past Microsoft was the low cost solution and Microsoft was then competing and attacking expensive proprietary systems from below. Now for the first time the tables are turned and it's Microsoft that's being attacked from below by a lower price solution.

—Brad Silverberg, former member of the Microsoft executive team (www.milestone-group.com/news/04_07/Silverberg.html)

There's no good DRM [digital rights management], period.

—Jean Bedord, publishing industry analyst at Shore Communications, as quoted in “Have e-books turned a page?” by David Becker on CNET News.com (news.com.com/Have+e-books+turned+a+page%3F/2100-1025_3-5326015.html?tag=st.pop)

Without open standards, software freedom is an illusion. Open standards are the foundation for software freedom.

—Larry Rosen, speaking at Open Source, Open Standards, heard live

After almost a year of rivalry, pmdisk and swsusp will be re-unified into swsusp, and Pavel Machek will continue to lead swsusp development, with help from Patrick Mochel who did the fork and other folks interested in the software suspend feature. Patrick made a nice apology to Pavel for forking the code, and the two have been working very nicely together since then, sharing patches and conversing on the technical issues. It's nice to see this dispute resolved, because it all emerged from a shared passion to make the code as good as it could be, to provide a solid software suspend feature and to merge the work into the official kernel with as little breakage as possible.

cryptoloop is very likely to be dropped from 2.6 unless Andrew Morton decides to wait to remove the code in 2.7; either way, it seems clear that cryptoloop is on the way out. The feature, intended to allow the user to mount encrypted filesystems over loopback, is apparently buggy and unmaintained, as well as having significant security problems. And as far as Andrew is concerned, it is this security question that dooms cryptoloop. Better to have no feature, he says, than a feature that falsely claims to provide real security. The argument, put forward by a number of developers, that such major changes as removing a feature should never occur in a stable series, is being heard, but it seems that this may only delay the inevitable. Unless someone steps up to maintain it and fix the security holes, cryptoloop probably will not last much longer.

The Reiser4 filesystem, a significant departure from Reiser3, has been accepted into the -mm kernel series (Andrew Morton's personal tree) and appears to be on the fast-track for inclusion in an official release. The Reiser folks are ecstatic about this, as it tremendously widens the field of users and thus, of bug reports. Reiser4, with only the core functionality submitted to Andrew, claims to be the fastest filesystem in the Linux world. That claim may involve quite a moving target, but undoubtedly this new generation of filesystems is picking up speed in general, which makes a huge difference to anyone involved in disk-intensive operations.

Andrew Morton, as the 2.6 maintainer, has decided to question some recent traditions, including the validity of having a stable and unstable kernel series. Although his innovations may not be so extensive as to discard the entire idea, he has stated that he feels it is in the realm of the distributions to bring true stability to the kernel. The official kernel sources, he says, should focus on speed and features as well as stability, but not give stability the high pinnacle of importance that it has been given for the past several major releases. There are many ways to think about this, and it's important to remember that Andrew, Linus and the rest always are modifying their development model in response to their own ideas and the ideas of others. Perhaps the most optimistic way to look at Andrew's current idea is as a way to expand the field of kernel development to include the distributions as full-class members of the development process, with stability as one of their primary aims.

DevFS, having been at the center of one of the largest controversies in Linux history, seems to be edging closer and closer toward finally leaving the kernel. Andrew Morton has said that the 2.8 kernel certainly will not have DevFS, and that the feature could be removed as early as mid-2005. There always has been a push to keep DevFS in until its replacement, udev, could provide all the features DevFS provided. For all its faults, DevFS has proven to be a tough feature to replace. Richard Gooch put a tremendous amount of work into it, and in spite of all the criticism he got from Alexander Viro and others, his work still has not been replaced, almost two years after Richard abandoned the project and gave up on kernel development entirely.

If you want to read more from Linux Journal authors, add our Web site to your bookmarks or RSS reader and catch their Web articles and columns.

The Battle for Wesnoth is a 2-D, turn-based game, with a sword-and-sorcery storyline, slick graphics and music and a lot of levels. Your units gain experience points and power, if you can keep them alive from battle to battle. Occupying villages gives you the ability to heal units and recruit more. Levels seem to make an abrupt jump in difficulty from fighting a few Orcs with allies and an umpteenth-level sorcerer on your side up to trying to defeat two armies of the undead with a handful of footpads and some fish-boys, so it's a good thing you don't have to be good to have some fun with this game.