Leading items

The Grumpy Editor's Obviously Incorrect 2006 Predictions

Your editor doesn't really know any more about what will happen this year than anybody else. But he has never been one to let such a difficulty stop him from embarrassing himself by posting predictions in front of thousands of people. So, without further ado, here's a set of highly unlikely prognostications for the coming year. As usual, no warranty applies.

Legal issues

The GPLv3 process will dominate the news for the first half of the year. The FSF seems fully aware of the stakes involved in a new version of the GPL, and Eben Moglen is the ideal person to push this effort forward. But there is no way that changes to such a fundamental document could be anything but controversial. How the FSF handles the feedback it gets will determine whether the resulting license is widely respected - and used.

The non-free kernel module issue will come to a head this year. Patience with these modules has been fading for years, while concern over the lack of free drivers for certain types of hardware is on the increase. This year, some developer or other is likely to force the issue and mount a more direct challenge to the legality of proprietary kernel modules. Others, meanwhile, will continue to make them harder to write and maintain. Either way, we may reach a point where the maintenance of, for example, proprietary drivers for video cards is no longer feasible. Whether the end result is the release of free drivers or the complete withdrawal of support remains to be seen.

The broadcast flag will be back, and European software patents too. The interests behind this sort of legislation never give up, so we'll never be able to stop fighting. But if we keep up the battle, we stand a respectable chance of winning much of the time.

Development

2006 will be the year of Linux on the desktop. Just like the last ten years. Don't expect any amazing advances, just slow, steady progress. The applications will get better, and people will slowly see more reasons to run Linux. Governmental mandates for open document formats - likely to proceed despite the tactics used in Massachusetts - will help in this regard.

The world will begin to discover alternatives to OpenOffice.org. OpenOffice is great stuff, and it lets Office workers move over to free software without overly disrupting their world. But there is a great deal of interesting work being done on platforms like AbiWord, KOffice, Gnumeric, etc. Once people get past "looks like Office" and start to concern themselves with issues like memory footprint or innovative new features, they will become more open to alternatives. Luckily for us, the free software community is strong enough to be able to provide those alternatives.

De-bloating will gain on new features as a development priority in many projects. This work will be driven partly by a general unease with the size of our systems, and partly by the increase in the number of developing-country hackers who are particularly motivated to make things run well on older, less capable hardware.

Perl 6 will not be released; it may not even be completely specified by the end of the year. We will, however, start seeing Perl 5 releases with more backported Perl 6 features.

The Fedora project will have to make changes to preserve developer and user interest in 2006. Fedora is still hard to contribute to, its decision process is relatively opaque, the promised Fedora Foundation is missing, the short support period keeps users on an upgrade treadmill, Fedora Legacy is not staffed at a level where it can be relied upon, and, crucially, other free, leading-edge distributions (OpenSUSE, Ubuntu) are increasingly competing for the same users. Fedora remains a top-quality distribution, but it risks losing some of the user and developer energy which makes it an important distribution.

Debian 'etch' will be released in December, on schedule -- or, at least, very close to it. The Debian developers are tired of their reputation for unreliable release schedules and see an opportunity to improve the situation.

Emacs 22 will be released. This prediction may seem like more of a stretch than even the Debian release, but the time is coming for the emacs hackers to show the world that they have not been idle all these years.

The pace of kernel development will not slow. The increased emphasis placed on avoiding regressions and user-space breakage will continue, however, and the quality of kernel releases will continue to go up. The kernel available one year from now will be substantially different from the current 2.6.15 release - but it will be good stuff.

Commerce

There will be an increasing number of Linux-based gadgets available. Embedded Linux is finally reaching the potential it has shown for many years, and it will show up in no end of interesting new toys. Unfortunately, most of those toys will be locked down and not hackable.

Novell will get its act together and become a truly successful Linux-based company. This result will be a combination of long experience in selling to large businesses, clueful people on staff, and a strong desire among customers to have more than one vendor to choose from.

Ubuntu/Canonical will start to make some real money. At some point the company has to bring in some revenue if it is to be sustainable over the long term. But, more to the point, the Ubuntu folks seem to be doing many things right: generating interest in the user and developer communities while pursuing goals (such as application certification) which make large customers happy.

Miscellaneous

iPod users will begin to notice two free operating systems for their toys, being iPodLinux and, toward the end of the year, Rockbox. The latter should be especially interesting to blind users, thanks to its voice menu feature. The advantages of free software for gadgets will become clear to more people - but so will the conflict with DRM schemes.

A Firefox vulnerability will be used to compromise systems. Firefox is too big and complex to be without vulnerabilities, and it is becoming too popular to ignore.

The SCO case will drag on, perhaps severely reduced by renewed motions from IBM and Novell. But few people will care anymore.

The safest prediction of all, of course, is that Linux and free software will continue to improve. The development momentum behind the free software community is truly amazing, and it shows no signs of slowing down. Whatever else happens over the next year, our systems will be stronger and more fun to work with. Your editor is looking forward to it.

Comments (51 posted)

GStreamer's MP3 for Linux

The MP3 audio format is a pain. It is patent-encumbered, making it hard for Linux distributors to package (or Linux users to use) legally in various parts of the world. It doesn't even sound all that good, compared to some of the alternatives. Yet MP3 is hard to avoid; digital audio players often prefer it, and much of the interesting audio content to be found on the net is encoded as MP3. So Linux users who do any amount of audio listening with their systems generally end up with MP3 software on their systems even if their distributor refuses to include it.

The hassles of tracking down unofficial repositories, configuring a system to use those repositories, and installing MP3-capable software are something that many Linux users take in stride. Using Linux has often required some of that kind of work, after all; you newcomers should just be happy that you don't have to come up with your own XFree86 modelines anymore. But the lack of native MP3 capability is an impediment for potential users who want things like audio to simply work without a bunch of fiddling around. Such people tend to be uninterested in discussions of the evil of software patents and the superiority of Vorbis audio. None of that helps them listen to their favorite Norwegian reggae Internet radio station.

The folks at Fluendo - the main force behind GStreamer - have made an attempt to improve this situation. Fluendo has bought a patent license for the MP3 technology, and has used it to make a couple of different items available:

• A GStreamer plugin for MP3 released under the BSD license; it is downloadable from the Fluendo site.

• A binary-only version of the plugin which has been made freely downloadable via the Fluendo web shop. The binary plugin is generated from the BSD-licensed source.

There are other freely-licensed MP3 decoders available, but the Fluendo release is still worthy of note due to its use of the BSD license. Most MP3 codecs are licensed under the GNU GPL, which includes this language:

This language makes it hard to distribute patent-encumbered, GPL-licensed code, so the redistribution of any application containing a GPL-licensed MP3 codec is legally dubious. The use of the BSD license is an attempt to avoid this particular clause of the GPL. One should not forget, however, that the result of linking BSD- and GPL-licensed code is normally a derivative work which must retain the GPL license. So it is not clear that any GPL-licensed GStreamer application can be shipped with the new decoder.

The binary-only plugin has a different intent. This plugin is fully licensed, so any Linux user (on a supported version of a supported distribution on a supported architecture) should be able to install it and play MP3 files without patent worries. Distributors can also sign a contract [PDF] with Fluendo which allows the binary plugin to be included with a distribution. There are plenty of restrictions in the contract, including an end-user license which prohibits further redistribution. So, while a distributor can gain the right to legally distribute the binary-only MP3 decoder, any derivative distributions must sign their own agreement with Fluendo to obtain the same right. The contract also prohibits "embedded" use, so projects like iPodLinux seem unlikely to be able to ship this plugin.

So Linux distributors can now ship MP3-capable distributions, as long as they don't mind the little fact that any such distribution is not 100% free. Which distributors will accept this deal remains to be seen; in a few cases, some guesses can be made based on the discussion (or lack thereof) on the relevant mailing lists:

• Fedora looks like it will sit for months waiting for Red Hat's lawyers to issue a pronouncement. One Red Hat staff member has said, however: "I expect there will be future developments in the media formats area of Fedora, but for now it seems unlikely that we will do more than link to the packages you have kindly made available."

• Mandriva already includes MP3 support in its distribution. Assuming that Mandriva feels safe in shipping MP3 codecs now, it probably sees little to gain by adding a binary-only version.

• Ubuntu seems to be considering including the plugin in its "restricted" area. There is some discussion of whether it would still be legal to include rhythmbox, which is GPL-licensed with no plugin exception, in such a work.

Most other major distributions do not currently appear to have a public discussion going.

In an ideal world, our systems would include free codecs for all of the widely-used audio and video formats. The world we actually live in, unfortunately, requires that we set our expectations a little lower. While many of us can do nicely with formats like Ogg much of the time, the simple fact is that missing MP3 support makes Linux less useful for many people. And this is not a problem that can be solved by coding. The contributions from Fluendo do not qualify as a solution, but they could well help make Linux work for people who were not able to do what they wanted previously. That's a step in the right direction, even if it is not ideal.

Comments (34 posted)

A proposed SonyBMG settlement

Just before the end of 2005, word got out that SonyBMG had put together a proposed settlement in of of the class-action suits spawned by its ill-advised copy protection measures. The EFF promptly signed on to the settlement as well. The full text is available in PDF format; the following is a summary.

The ostensible plaintiffs in this action - SonyBMG customers who installed the DRM software found on SonyBMG discs - don't get a whole lot directly. The settlement allows for XCP victims to get a non-DRM version of their discs, to download MP3 copies of the songs on the discs, the right to download one album "from a list of more than 200 titles," and the option of (1) three more album downloads or (2) a check for $7.50. People who bought MediaMax-protected discs only get one album download.

That is not a whole lot of compensation for somebody whose computer has been compromised by SonyBMG's malware. It rather differs from the hard line taken by the recording studios against those deemed to be "pirates." This result is not particularly surprising, however; class-action suits are rarely about the interests of the people named as plaintiffs. Nonetheless, there are provisions of this settlement which will benefit those plaintiffs - and many others as well. They include:

• SonyBMG agrees to immediately recall all CDs containing the XCP software. In theory, this recall has already happened, but there have been numerous reports of XCP discs remaining on store shelves.

• The company will also stop manufacturing CDs with the MediaMax DRM software - for at least two years. MediaMax is not quite as bad as XCP, but it still has "phone home" capabilities and can open up a system to security problems.

• SonyBMG will provide uninstallers for XCP and MediaMax, and a security update for MediaMax as well.

• Numerous behavioral changes are called for; SonyBMG agrees not to install software without a positive agreement, to make uninstallers available, to describe the functionality of software to the user "in plain English," to, refrain from collecting data on users, to issue patches for security problems, and to "obtain comments" on its EULAs and potential security vulnerabilities in its future DRM software. These constraints only apply through 2007, however.

Together, these terms comprise a set of rules that music distributors might be expected to play by in the future. On the good side, they call for explicit information on what DRM software does, limitations on phoning home, the availability of uninstallers, and some attention to security issues. That's a start, and more than was available before.

On the other hand, this settlement fails to address fundamental questions, such as whether it is right to force people to install software to listen to music they have purchased. Limitations on fair use, including making backup copies or putting music on a portable player, are not addressed. This settlement makes it clear that DRM software does not have complete freedom on the user's computer, but it in no way questions the correctness of that software in the first place. The entertainment industry remains free to make its DRM regime is restrictive as it likes, as long as it does not step on users' toes in other ways.

In other words, SonyBMG's original purpose for XCP - keeping its customers from putting music onto their iPods - has not been addressed. The company is free to attempt to impose the same restrictions in the future. The people behind the suit can claim a win, and the lawyers will certainly get their (currently unspecified) cut. The court will likely approve the settlement, but SonyBMG is not out of the woods yet. Various other lawsuits are still outstanding, including one in Texas which alleges spyware violations.

Why the EFF signed on to this agreement is not entirely clear; perhaps declaring victory was more important trying to fight the larger battle. It would have been nice if this case could have been used to attack the assumptions and goals behind DRM in general, rather than being satisfied with the creation of a basic DRM code of conduct. That is a battle which will have to be fought another day.

Comments (5 posted)

Novell's XGL code posted

The December 22 LWN Weekly Edition contains an article on how a significant amount of XGL (X over OpenGL) work has been done behind closed doors at Novell. XGL hacker David Reveman has now posted Novell's code with a request that it be added to the freedesktop.org CVS repository. Large amounts of work have been done; see David's mail for the summary. Now all that work has to be somehow merged with what the rest of the XGL developers have been doing in the open; this may turn out to be a long process.

Comments (2 posted)