UpFront

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

“It's in my old mail somewhere” isn't good enough. Add a powerful search feature to your mail without switching mailers with this simple command-line tool. You can search for any text or for words that appear in certain headers.

To use Mairix, edit a short config file to specify where your mail folders live, then run it without arguments to build the index. Run Mairix with an item to search on, and it puts copies of search results in a match folder that you can browse in your mailer like any other mail folder.

This real-time strategy game breaks away from the Warcraft style, where you select units and give orders. Instead, you create a building site or other task and specify how many of your amoeba-like units you want to work on it. Based on units you have available and their skill levels, the game will assign units and tasks.

There's a beautiful level editor that creates interesting random terrain, but the computer players are still fairly dumb. Globulation 2 supports network play, and development seems to be happening rapidly.

People's stereotype [of the typical Linux developer] is of a male computer geek working in his basement writing code in his spare time, purely for the love of his craft. Such people were a significant force up until about five years ago.

—Andrew Morton, gcn.com/vol1_no1/daily-updates/26641-1.html

Openness matters in two places. One is in source code. The other is in data.

—R0ml Lefkowitz, Director, Open Source, AT&T Wireless (Talk at the O'Reilly Open Source Convention)

Never before in history have we been able to see incumbent businesses protect business models based on old technology against creative destruction by new technologies. And they're doing it by manipulating the political process. The telegraph didn't prevent the telephone, the railroad didn't prevent the automobile. But now, because of the immense amounts of money that they're spending on lobbying and the need for immense amounts of money for media, the political process is being manipulated by incumbents.

—Howard Rheingold, www.businessweek.com/bwdaily/dnflash/aug2004/nf20040811_1095_db_81.htm

This adoption of free software to resolve incompatibility between the economic need for provider diversity and the engineering need to avoid product diversity is, I think, fairly unique across all industry. I can't think of similar examples.

—Andrew Morton, www.groklaw.net/article.php?story=20040802115731932

Red Hat has decided to release their newly purchased Global Filesystem (GFS) clustering filesystem under the GPL. This project has had a checkered past. It started out as a GPLed project from Sistina, but the company changed its license in 2001 to the Sistina Public License, which required a licensing fee to be paid to Sistina in the event of source code redistribution. Among various outcries at that time, Alan Cox claimed the license change violated the copyright of his own GFS contributions, and the OpenGFS Project sprang up, using the last GPLed version of Sistina's code. During the next few years, Sistina made some effort to market GFS in its new proprietary form, including partnering with CommVault in 2003. In 2004, Red Hat bought the code from Sistina and has now re-released it under the GPL. Back in 2000, GFS was considered a likely candidate for inclusion in the official Linux kernel; now that it is available again, Red Hat is keen to submit it for inclusion once more. This time, it looks as though the code will stay free.

Linus Torvalds has proposed a new patch attribution convention, which looks to be achieving an early success. The goal, as he puts it, is to enable kernel developers to track the history of a patch, in the case of charges of copyright violation. There's no confusion over Linus' inspiration here. He and others have had to spend a lot of time debunking each of The SCO Group's charges of copyright violation. Doing so has, until now, involved much wading through ancient mailing-list archives. Linus' suggested attribution system, which appears to be being adopted fairly widely after only a brief discussion, simply involves developers including their names on patches to indicate compliance with the kernel's license. Each patch will have one name for each developer who edits it before it is included in the kernel. This way, any future charges of copyright violation can be investigated by examining the patches themselves, rather than trying to trace each patch's history through mailing-list discussions. As of this writing, many developers are using the system, although so far there has been no need to use the newly gathered data.

Randy Dunlap has introduced a new wrinkle into the old idea of saving .config information in the kernel itself after compilation. His idea is also to include data on the kernel version, as well as the date compilation took place. The initial struggle to get .config information into the kernel was fraught with controversy, due to the argument that such data also could be stored outside the kernel. With that part of the controversy resolved, the suggested addition of Randy's new data is finding a much warmer reception. In fact, a number of developers have remarked that the idea itself should have been obvious long before. It seems a deeply embedded part of human nature that even the obvious often must have its initial discoverer.

Jeff Dike's User-Mode Linux (UML) is having some technical difficulties making its way into the official 2.6 kernel tree. Apparently, Andrew Morton is more than happy to accept Jeff's patches, although Jeff appears to be having trouble splitting the patches into acceptable chunks. Jeff himself has said that he has “painted himself into a corner” with regard to his UML work; the problem, he says, is finding the right tools to manage the patch split. The UML patch has become so large that splitting it up has become a major undertaking. This actually is a typical occurrence with large features. Often, insufficient effort is given toward preparing the patch for inclusion, and when the developers finally feel the time is right, they discover they have a ton of additional work to do before the patch even will be considered. This usually engenders much controversy. Jeff is no stranger to the issues involved, but even knowing the requirements, it is still difficult to split patches into atomic pieces that each either fix or implement a single thing. Clearly, UML is destined for inclusion in 2.6, but these difficulties may result in significant delays.

John A. Martin has maintained the CREDITS file for quite a long time and has been acknowledged in the MAINTAINERS file for that reason as well; but the CREDITS file now apparently has become self-maintaining, no longer requiring any specific maintainer. Linus Torvalds, Andrew Morton and the other kernel maintainers have taken over much of that role, and developer patches often include their own updates to the CREDITS file, without requiring anyone else to add them. The CREDITS file has, quite simply, become a fully adopted element of kernel development. Back in the old days, when that file was first conceived, the task was much more daunting, because there were so many contributors not mentioned in it. Now that it has established itself, its current listing is much more accurate. John graciously stepped down when Adrian Bunk pointed out that a maintainer was no longer needed; although John said he would be willing to resume maintenance should the need arise in the future.