GStreamer's MP3 for Linux: Thanks, but no thanks. [LWN.net]

GStreamer's MP3 for Linux: Thanks, but no thanks.

Posted Jan 2, 2006 22:42 UTC (Mon) by jdub (subscriber, #27)
In reply to: GStreamer's MP3 for Linux: Thanks, but no thanks. by spot
Parent article: GStreamer's MP3 for Linux

Well, firstly - this is the patent-license for the binary, not the source. The code is MIT licensed, and thus, 100% Open Source. Sure, Fluendo can't grant permission to distribute binaries based on random use of the MIT code - you must enter into a contract with Fluendo to receive a patent license.

MP3 is hard. Patent licenses are hard. Here, at least, is a legal option for those who care about solving the problem legally, and with fairly minimal intrusion for users and distributors... *if* reality means more to you than some of the more abstract notions of software freedom (ie. if you were being serious about software freedom, you wouldn't be using MP3 in the first place).

MAD has a serious problem: Look at its license (GPL), then read section 7 of it. Bad news for everyone.

(Log in to post comments)

GStreamer's MP3 for Linux: Thanks, but no thanks.

Posted Jan 2, 2006 22:47 UTC (Mon) by piman (subscriber, #8957) [Link]

> MAD has a serious problem: Look at its license (GPL), then read section 7 of it. Bad news for everyone.

Of course, most MP3-playing software is also under the GPL. Most useful software on GNU/Linux is under the GPL, period. So regardless of this plugin's license, if you distribute this plugin with software the end result is probably going to be under the GPL.

Jon mentioned this in the article -- Rhythmbox has no license exception yet, and I'm fairly sure Muine doesn't either. I'm less sure about amaroK, but at least its debian/copyright file doesn't have an exception.

So you can't actually ship the plugin with any of the major GStreamer-based music players. It solves nothing.

GStreamer's MP3 for Linux: Thanks, but no thanks.

Posted Jan 2, 2006 23:07 UTC (Mon) by jdub (subscriber, #27) [Link]

Sure, no argument: You can't distribute the plugin with a GPL user of it. That puts the responsibility on users - and they can do whatever they wish, because they're not distributing it. So even with GPL software on top, the plugin has a good use case, which it solves legally and well.

MAD being licensed under the GPL is another story: It is incompatible with itself and not distributable at all. Read section 7.

GStreamer's MP3 for Linux: Thanks, but no thanks.

Posted Jan 3, 2006 1:19 UTC (Tue) by JoeBuck (subscriber, #2330) [Link]

Section 7 of the GPL means that people in countries where the MP3 patents apply to software (for example, the USA) cannot distribute GPL MP3 code. However, it does not constrain those in countries where software patents are not yet valid. This means that, for now, sites like livna do not violate the GPL, even if they allow Americans to access them, and Americans downloading MP3 support from sites like livna aren't violating the GPL, though they may be infringing patents by using the code.

In practice, though, only the copyright holder can enforce a GPL violation, and they won't in this case, so the only danger is from the patent infringement. And that applies to MP3 code regardless of license.

GStreamer's MP3 for Linux: Thanks, but no thanks.

Posted Jan 11, 2006 20:00 UTC (Wed) by pimlott (subscriber, #1535) [Link]

Section 7 of the GPL means that people in countries where the MP3 patents apply to software (for example, the USA) cannot distribute GPL MP3 code.

Interpretations of Section 7 have gotten mad! This statement is as misleading as

Section 7 of the GPL means that you cannot distribute the code you stole by hacking into Microsoft under the GPL.

Neither is legal, but not because of Section 7. Indeed, the GPL itself considers Section 7 a no-op:

This section is intended to make thoroughly clear what is believed to be a consequence of the rest of this License.

A summary of Section 7 would be:

If you cannot distribute a body of code under the GPL, you cannot distribute the code under the GPL.

A shorter summary would be:

True.

I'm being extreme--probably Section 7 has some legal significance. However, I don't believe that it forbids distribution in any cases where it would otherwise be allowed. I know that Eben Moglen has said or suggested otherwise, but I haven't heard a plausible explanation. And to be clear, I consider the "for example" in Section 7 a complete non-sequitur.

reality & freedom

Posted Jan 3, 2006 1:39 UTC (Tue) by pjm (subscriber, #2080) [Link]

The previous poster suggests that the availability of this binary is clearly a win in real terms, but this is far from clear in the longer term.

It would clearly be contradictory to think that software freedoms have no connection with reality: each of the freedoms in question is in the physical world, reality.

Inability to share this binary among others is one real-world result. Other real-world harms come from perpetuating the idea that it is wrong to help one's neighbour by sharing one's software.

Inability to modify the source and use the resulting different binary has yet other harms: for example it is harder to take advantage of bug fixes & feature improvements of others.

Additional convenience in using the mp3 format reduces the rate of transition to better formats. Thus, Fluendo might instead consider distributing a command-line utility to convert mp3 files to better formats (allowing separate, unencumbered, gui interfaces to that tool). This would have more chance of helping the transition to other formats; whereas a gstreamer plugin will clearly reduce this transition.

Paying the patent license fee and slowing down the transition to better formats each increase monetary payments to the patent holder, and increase the attractiveness of patenting things (to the detriment of everyone else), increasing the likelihood of future patent harm.

Of course this binary does provide real-world benefit as well: one immediate benefit of easing the playing of some music files, and future benefits through increasing the use of other free software for people who would use mp3 software no matter what.

reality & freedom

Posted Jan 3, 2006 15:30 UTC (Tue) by Felix.Braun (subscriber, #3032) [Link]

Can Fluendo also write a program that lets me play vorbis files on my portable media player? Didn't think so.

Of course, people who care about software freedom could vote with their euros and buy media players that do play vorbis. But try telling that to the novice computer users that some linux distributions are trying to attract. To these people software freedom looks like an inhibition to use gadgets that run quite well under competing setups.

So, by lowering the inhibitions to such users, Fluendo is arguably helping Linux adoption by a wider public. Which might even help convince hardware vendors to support alternative audio formats in their firmware.

GStreamer's MP3 for Linux: Thanks, but no thanks.

Posted Jan 3, 2006 13:06 UTC (Tue) by rickmoen (subscriber, #6943) [Link]

jdub wrote:

The code is MIT licensed, and thus, 100% Open Source.

Sorry, Jeff; that doesn't follow. Being under an OSI-approved licence is necessary but not sufficient to make the codebase open source. Additionally, source code must be actually available (which is true in this case, but needn't be the case for somethine merely declared to be "MIT licensed"), and there must not be external legal obstacles preventing exercise of the rights in that OSI-approved licence.

Such as, notably, patents.

Thus, in parts of the world where such a codebase is patent-encumbered (rights not exercisable on account of patent-based restrictions), it's proprietary by definition, as long as those conditions persist.

That's not to say I might not, in some cases, ignore a patent I felt to be either meritless or basically harmless. But it's simply wrong to say that the licence terms are all that matter. No, it's those terms and the ability to exercise them that matter.

Rick Moen
rick@linuxmafia.com

GStreamer's MP3 for Linux: Thanks, but no thanks.

Posted Jan 3, 2006 16:48 UTC (Tue) by ranger (guest, #6415) [Link]

Such as, notably, patents.
Thus, in parts of the world where such a codebase is patent-encumbered (rights not exercisable on account of patent-based restrictions), it's proprietary by definition, as long as those conditions persist.

By your definition, the Linux kernel is proprietary, since it includes patent-encumbered code (such as RCU).

But, the Linux kernel is not proprietary, and considered Open Source (and Free Software etc etc).

Why?

Because of the license IBM has granted for the use of the RCU code, which most people agree has acceptable conditions.

So, the real issue is whether the Linux distribution in question is agreeable with any license terms on an MP3 decoding license.

For example, Mandriva has negotiated with Thompson that it (and others using any Mandriva-based distribution for a non-embedded device) may provide royalty-free MP3-decoding, so MP3 playback works out-the-box.

While it would be nice to have a policy on this in each Linux distribution, people are still arguing the wrong points in this thread.

If you don't want any patent-encumbered code, better not run a 2.6 kernel.

GStreamer's MP3 for Linux: Thanks, but no thanks.

Posted Jan 3, 2006 17:07 UTC (Tue) by rickmoen (subscriber, #6943) [Link]

"ranger" wrote:

But, the Linux kernel is not proprietary, and considered Open Source (and Free Software etc etc). Why? Because of the license IBM has granted for the use of the RCU code, which most people agree has acceptable conditions.

You seem unaware that you are actually in strenous agreement with me -- presumably because you didn't read what I wrote attentively.

Allow me to reiterate: "patent-encumbered (rights not exercisable on account of patent-based restrictions)" and "external legal obstacles preventing exercise of the rights".

I'm extremely well aware that by no means does the mere existence of an applicable patent (or other legal encumbrance) does not, in itself, preclude exercise of open-source licensing rights. My comments concerned particular patent encumbrances (and other legal obstances) that do happen to be imposed in a fashion that blocks exercise of those rights.

You, unfortunately, didn't notice that I'd already carefully observed that (elementary) distinction in my post, which is why I'm now obliged to waste time repeating myself. Oh well.

Rick Moen
rick@linuxmafia.com