Cisco to release an open-source H.264 codec [LWN.net]

Cisco to release an "open-source" binary-only H.264 codec

Posted Oct 30, 2013 13:35 UTC (Wed) by pizza (subscriber, #46) [Link] (2 responses)

The announcement says "We plan to open-source our H.264 codec, and to provide it as a binary module that can be downloaded for free from the Internet." The various quotes from Mozilla all reference the binary blobs; There's no mention of any source code release.

So I'm confused; if they're just providing binaries, how can they call this "open source"?

(That said, even if it's just binaries, kudos to Cisco for essentially guaranteeing that nobody will ever have to pay license fees for a software implementation of h.264..)

Cisco to release an "open-source" binary-only H.264 codec

Posted Oct 30, 2013 13:39 UTC (Wed) by corbet (editor, #1) [Link]

My understanding is that they will provide source. But if you use that source directly, you're on your own with regard to the patent issues. If you get the binary blob, instead, Cisco nicely picks up the patent royalties.

It is a solution quite similar to the famous Fluendo MP3 decoder. Not ideal by any means, but it allows the sidestepping of some patent hassles.

Cisco to release an "open-source" binary-only H.264 codec

Posted Oct 30, 2013 17:29 UTC (Wed) by b7j0c (guest, #27559) [Link]

don't offer cisco any kudos. web video is still a young enterprise at least as far as sales go relative to non-web video. cisco is merely extending the "don't ask, don't tell" agenda of the mpeg-la to the few remaining holdouts before sales of web video start to eclipse other online channels.

at some point in the future the mpeg-la will remind us of the fine print.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 30, 2013 13:38 UTC (Wed) by johnsmith_notthatone (guest, #92971) [Link] (66 responses)

Money quote:

«Here’s a little more detail about how things are going to work: Cisco is going to release, under the BSD license, an H.264 stack, and build it into binary modules compiled for all popular or feasibly supportable platforms, which can be loaded into any application (including Firefox). The binary modules will be available for download from Cisco, and Cisco will pay for the patent license from the MPEG LA. Firefox will automatically download and install the appropriate binary module onto each user’s machine when needed, unless disabled in the user’s preferences.»

And see how Mozilla shows its true colours:

«While Cisco’s move helps add H.264 support to Firefox on all OSes, we will continue to support VP8, both for the HTML video element and for WebRTC. VP8 and H.264 are both good codecs for WebRTC, and we believe that at this point, users are best served by having both choices.»

So, now a patent encumbered and binary only module without an audio part is good enough! How the righteous have fallen!

Because the audio part is missing:

«Provided we can get AAC audio decoders to match, using Cisco’s OpenH264 binary modules allows us to extend support to other platforms and uses of H.264.»

With the appropriate hypocrisy from Cisco:

«Cisco has a long-standing history of supporting and integrating open standards, open formats and open source technologies as a model for delivering greater flexibility and interoperability to users. We look forward to collaborating with Mozilla to help bring H.264
to the Web and to the Internet.»

And that includes all the freedom-loving action called "censorship"!

After engaging with the privacy destroyer "Facebook", now a collaboration with the censorship promoter "Cisco".

What's next, Mozilla, what's next? Where were all the principles gone?

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 30, 2013 14:22 UTC (Wed) by lambda (subscriber, #40735) [Link] (23 responses)

Oh, come on. Mozilla is one of the largest backers of end-user free software and open standards. They have fought the good fight for patent-free codecs, but the fact is that they are just not viable in the marketplace. H.264 is better than the patent-free codecs that were available at the time the <video> element was originally designed (at the time, the best choice was Theora, which was considerably worse). VP8 was too little too late; it's not really any better than H.264, and wasn't released as open until H.264 had already gained years of traction and hardware implementation.

Mozilla has realized, pragmatically, that this fight is over. All it will do by not supporting H.264 is make the experience worse for its end users, and delay the adoption of open standards like WebRTC due to compatibility issues. They had already decided to use platform-supplied H.264 codecs if available; this simply allows them to have one that is available even if there isn't a platform-supplied H.264 codec.

As far as end-user freedom is concerned, this is just as free as anything else. You can download the source and build it yourself. The only issue is if you want to redistribute your rebuilt sources, at which point you run into the patent license issue. That doesn't apply in countries which don't recognize software patents, this doesn't apply once the patent licenses expire, and this doesn't apply if you simply ignore the patent license issue, which many free software users choose to do by downloading and installing restricted codecs even in countries where those codecs are technically patent encumbered. Are you telling me that you don't have an MP3 codec installed? That you went without a GIF decoder while those patents were still in effect?

Mozilla has realized that the fight over the current generation codec is over; H.264 has won, and failing to support it simply hurts their users for no real benefit. But they have redoubled their efforts for the next generation; rather than hoping passively that someone will come along and produce a free codec, they have now hired several of the developers of Daala, a next generation codec that looks like it may have a chance of being significantly technically superior to H.265.

While you may not agree with all of the compromises they have chosen to make, please do recognize the contributions Mozilla has made to free software, open, unencumbered standards, user privacy, and the like. And if you don't like their choice, what they're distributing is free software. Feel free to fork it and remove the parts you don't like.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 30, 2013 17:38 UTC (Wed) by b7j0c (guest, #27559) [Link] (6 responses)

how can people say open codecs have "lost"? web video has barely begun. non-web sales of video still dwarf web-hosted solutions and will for the foreseeable future. was flash unkillable? no. when the ipad was introduced, 98% of browsers had flash installed. yet flash was essentially slain. it will take a decade or more for bandwidth in the US to allow the majority of video content viewed by consumers to be web-based. we had of plenty time to build free alternatives....well, until mozilla caved

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 30, 2013 18:11 UTC (Wed) by khim (subscriber, #9252) [Link] (5 responses)

Wow. So many contradictions in so few words.

was flash unkillable? no. when the ipad was introduced, 98% of browsers had flash installed. yet flash was essentially slain.

By Apple, not by FOSS.

it will take a decade or more for bandwidth in the US to allow the majority of video content viewed by consumers to be web-based. we had of plenty time to build free alternatives....

Really? In this kind of world? Sorry to disappoint you, but that battle is already lost. Video is common enough on web that any platform which can not show it will not survive. And most of the video is delivered as H.264 nowadays.

how can people say open codecs have "lost"?

Because most users in US will never be able to use them? iOS devices don't support VP8 and it's unlikely that said support will ever materialize. Windows 8 can support VP8, but it's not available "out of the box".

You are all too ready to celebrate the death of Flash which Apple accomplished but somehow use this very example to explain that Apple could not repeat the feat with VP8 which is many times weaker position. This is… just crazy.

we had of plenty time to build free alternatives....well, until mozilla caved

Well, we could have seen the world where Mozilla would have joined it predecessor (Netscape Navigator) but instead it'll be popular browser for a few more years. Of course to actually remain popular it'll need to implement EME soon and after last year I can just imagine the amount of howling on LWN and elsewhere when that'll happen.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 30, 2013 21:12 UTC (Wed) by b7j0c (guest, #27559) [Link] (4 responses)

"And most of the video is delivered as H.264 nowadays."

no, 95% of video is delivered by cable, and it would not be possible to push this content over ip networks today

but one day this will change, but not today or even in five years. web video is still in its infancy relative to total media consumption.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 30, 2013 21:25 UTC (Wed) by lordsutch (guest, #53) [Link] (1 responses)

And most of that content on cable is (as of 2013) compressed with either MPEG-2 or some variant of MPEG-4 (including AVC, aka H.264).

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 30, 2013 21:37 UTC (Wed) by cesarb (subscriber, #6266) [Link]

Over-the-air digital television here in Brazil (SBTVD) uses H.264 too.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 30, 2013 22:33 UTC (Wed) by khim (subscriber, #9252) [Link] (1 responses)

no, 95% of video is delivered by cable, and it would not be possible to push this content over ip networks today

What are you talking about? Video is delivered quite often today via ip networks. Sure they are using multicasting, but it's still the same IP and they are using H.222 or H.264 almost exclusively.

I'm not even sure when I've seen video not delivered in such a way. Probably between two years ago and five years ago (I could have seen video delivered in some other forms in some obscure corners of the world, but even there most video comes from satellites which means H.222 or H.264, although, admittedly, not IP).

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 31, 2013 8:41 UTC (Thu) by marcH (subscriber, #57642) [Link]

> What are you talking about? Video is delivered quite often today via ip networks.

I really doubt that the "IP versus cable" transport question is relevant in a codec discussion.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 30, 2013 18:11 UTC (Wed) by gmaxwell (guest, #30048) [Link] (15 responses)

> That doesn't apply in countries which don't recognize software patents

Precisely which countries are you speaking of here? Presumably one not represented on this list: http://www.mpegla.com/main/programs/avc/Documents/avc-att... ?

You need to be careful with "software patents". Software patents is a lawerese term of art for patents that refer to software without any reference of any physical embodiment. Post Bilski (or pre-state street) you can call the US a country which doesn't recognize software patents... and yet the US is a country (one of something like forty-six) that has patents deemed essential to H.264 by the MPEG-LA's "third party" analysis.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 30, 2013 20:43 UTC (Wed) by nybble41 (subscriber, #55106) [Link] (13 responses)

> Software patents is a lawerese term of art for patents that refer to software without any reference of any physical embodiment.

"Lawyerese" aside, software patents include any patent which can be infringed simply by writing software, distributing software, or running software on a general-purpose computer without a license. Any other definition is merely an attempt by the pro-software-patent opposition to define away the problem. These H.264 patents are definitely the sort of thing people mean when they talk about software patents, whether or not they refer to any "physical embodiment". If they weren't software patents, they wouldn't be any threat to developers or users of open-source implementations of H.264. Obviously that isn't the case.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 30, 2013 20:51 UTC (Wed) by gmaxwell (guest, #30048) [Link]

> Any other definition is

What lambda must have been referring to with "countries which don't recognize software patents", because— as the MPEG-LA H.264 patent list shows, patents covering H.264 exist all over the world.

The only way you can really talk about "countries which don't" on any scale is by invoking that useless lawyerese definition, which was my point.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 30, 2013 22:27 UTC (Wed) by khim (subscriber, #9252) [Link] (11 responses)

These H.264 patents are definitely the sort of thing people mean when they talk about software patents, whether or not they refer to any "physical embodiment".

Really? How come? I can see how software which is supposed to work on abstract mathematical objects can be treated as something not related to “physical embodiment”, but H.264? Nope. It's really easy to construct a video file which will be completely garbled by H.264. As in: it'll look nothing like original. Sure, for that to happen you need to carefully construct “really bad” file, but the fact is clear: H.264 deals specifically with videos which resemble real-world objects and which are viewed by actual human (again: it's relatively easy to create a device which will be able to catch difference between raw video and even super-HQ H.254-encoded video).

If you remove “physical embodiment” then algorithms included in H.254 suddenly stop making any sense! They don't deal with some abstract math, they deal with real-world objects—and this is central property of H.254! Even when there are some software models used to tweak some properties of H.254 they are not presented as abstract math constructs, but instead they represent some aspects of human eye and brain!

IOW: even if “pure” software patents will be declared null and void H.254 patents will probably stay. That's why they are so dangerous. They can not invalidated because they are software patents. They probably can be invalidated if they will be found “obvious if you include some previously-unknown prior art”, but it's not easy to repel patents using this approach.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 31, 2013 0:21 UTC (Thu) by dashesy (guest, #74652) [Link] (4 responses)

Even if H.264 is that close to human perception, it goes to the same category as human genes. It makes even less sense to patent human genes, a method to find a property yes (like a method to find the objects), but not the actual biological facts. Understanding the visual perception is like decoding the brain, with much older prior art, preceding humans.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 31, 2013 2:04 UTC (Thu) by khim (subscriber, #9252) [Link] (3 responses)

Even if H.264 is that close to human perception, it goes to the same category as human genes.

You mean the decision that a naturally occurring DNA segment is a product of nature and not patent eligible merely because it has been isolated, butsynthesized DNA sequences, not occurring in nature, can still be eligible for patent protection, right? Well, nature certainly does some compression in human brain, but I'm pretty sure it's not all that similar to what H.264 is doing. The goal may be similar, but H.264 was not produced by careful study of human brain, that's for sure.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 31, 2013 8:43 UTC (Thu) by marcH (subscriber, #57642) [Link] (2 responses)

> H.264 was not produced by careful study of human brain, that's for sure.

/me thought visual perception happened in the brain... where then? I'm afraid now!

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 31, 2013 10:15 UTC (Thu) by khim (subscriber, #9252) [Link] (1 responses)

Visual perception obviously happens in the brain, but H.264 models don't really dig this deep. They use only some easily observable facts about brain which can be obtained without slicing and dicing dozen of brains of living humans.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 31, 2013 15:01 UTC (Thu) by dashesy (guest, #74652) [Link]

Slicing and dicing is not the usual method to study human brain, specially the living ones, psychological tests, eye tracking, electromagnetic imaging, together with surface and deep electrodes, and stimulants is. Kind of like reverse engineering.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 31, 2013 0:40 UTC (Thu) by nybble41 (subscriber, #55106) [Link] (5 responses)

By your bizarre definition, there wouldn't be any such thing as a software patent. After all, the input for any software must comes from some physical hardware device, the output must be rendered on a different hardware device, and the software itself must be stored in some sort of memory. Under your definition a patent on a program to evaluate F=m*a would not be a software patent, simply because it describes the motion of physical objects.

H.264 is a transformation between two abstract mathematical objects, to use your terminology. One is raw video, a large multi-dimensional array of numbers, and the other is H.264 compressed video, a bitstream. There is nothing physical involved. Given sufficient time, you could implement H.264 in your head, with perhaps a pencil and some paper to supplement your memory. That's about as abstract as it gets.

Some algorithms are more useful for solving a given problem than others. Like all lossy compression schemes, H.264 makes assumptions about patterns in the input and what might be important at the output stage. These assumptions are based on physical measurements and influence the design of the algorithm, but they don't make the algorithm a physical process or give it a "physical embodiment". They fall under the category of "defining the problem". The solution, the patented algorithms, are abstract mathematical formulas.

As I said before, a software patent is a patent which can be infringed by writing, distributing, or executing software. What the algorithm is or how it's derived make no difference whatsoever. If encoding a video with x264 (without a license from MPEG LA) infringes on some patent, then that patent is a software patent.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 31, 2013 2:41 UTC (Thu) by khim (subscriber, #9252) [Link] (4 responses)

By your bizarre definition, there wouldn't be any such thing as a software patent.

Why not?

After all, the input for any software must comes from some physical hardware device, the output must be rendered on a different hardware device, and the software itself must be stored in some sort of memory.

And all that much happen on planet Earth in our Galaxy. What this has to do with anything? Patent describes what it describes. If patent is equally applicable to any conceivable input (think, e.g., RSA… or LZW) then you need to stretch “machine with software embedded” idea quite far till it'll cover this thing: it can be used to encrypt or compress any data in any galaxy. It may be more or less effective depending on some nuances of alien technology and physiology, but it'll work.

H.264 is not like that: it's easy to imagine some alien race which will notice difference between original and compressed version. And it's easy to see that other implementations can exist (think VP8). It's algorithms were invented by humans and then tested on objects from real world (some of these objects are humans, too, but that's not a problem: a lot of patented inventions are invented to work with humans).

Under your definition a patent on a program to evaluate F=m*a would not be a software patent, simply because it describes the motion of physical objects.

Not at all. The fact that F=m*a does not depend on human ingenuity, it's just physics law. You may try to argue that notation used in F=m*a deserves patent and that's certainly true, but it's kind of too late to patent it: it was invented centuries ago.

These assumptions are based on physical measurements and influence the design of the algorithm, but they don't make the algorithm a physical process or give it a "physical embodiment".

Obviously not. What makes this “physical embodiment” is the fact that properties of physical object are important for the patent. What makes it patentable is that they don't dictate the design. Think steam engine patents or loom patent. Physical properties of steam or textile limit the possibilities, but then human ingenuity is used to pick some implementation (which supposedly provide tangible benefits). Similarly with H.264: physical properties of video records and human eye define the “research field” but then human ingenuity is used to pick some approaches (and to reject other ones). If there are no relation of “research field” to real world then theoretically patent should not granted (that's “pure math”), if there are no “human ingenuity” (that is: something can be done in exactly one way and physics says that no other possibilities exist) then patent also should not be granted.

The solution, the patented algorithms, are abstract mathematical formulas.

No. Patented solution is “a machine” which takes raw input from the image sensor (well, not 100% raw: it's somewhat preprocessed), stores it in some more compact form on the piece of “spinning rust” (or some other similar medium) and then plays it back.

As I said before, a software patent is a patent which can be infringed by writing, distributing, or executing software.

Which nowadays includes pretty much any “traditional” patent (or perhaps you've not heard about 3D printing)?

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 31, 2013 18:11 UTC (Thu) by nybble41 (subscriber, #55106) [Link] (3 responses)

> If patent is equally applicable to any conceivable input (think, e.g., RSA… or LZW) then you need to stretch “machine with software embedded” idea quite far till it'll cover this thing: it can be used to encrypt or compress any data in any galaxy. It may be more or less effective depending on some nuances of alien technology and physiology, but it'll work.

That may be true of encryption, but not compression. There is no compression algorithm, lossy or lossless, which is effective (output smaller than input) for all inputs. That would mean you could recursively compress any input--losslessly--down to at most one bit, which is clearly absurd. Video codecs are just compression algorithms which happen to be more effective for video than for other inputs, just as LZW is _less_ effective for video than for other inputs. It all comes down to designing your compression algorithm around the patterns you expect to see in the input, and in the case of lossy algorithms, what data you care about vs. what you consider noise.

> What makes this “physical embodiment” is the fact that properties of physical object are important for the patent. What makes it patentable is that they don't dictate the design. Think steam engine patents or loom patent.

Steam engines and looms have "physical embodiments" because they're physical machines which transform physical materials from one form to another. The blueprints don't infringe the patents; the physical devices do. It has nothing whatsoever to do with the importance to the patent of properties of physical objects. (And software is really just the blueprint for an abstract mathematical evaluation process; the computer it's running on is only a catalyst to speed up the rote work of evaluation.)

> The fact that F=m*a does not depend on human ingenuity, it's just physics law.

The parameters of an effective video codec are a bit higher-level than inertia, to be sure, but the formula F=m*a and what information you can discard to improve your video compression ratio without noticeably affecting human perception are both approximations extrapolated from observed facts about the physical world.

F=m*a isn't a physical law. It's just one out of an infinite number of formulas which fit the observed facts (at least until you factor in relativity and quantum mechanics). The selection of this formula rather than one of myriad alternatives which fit the facts equally well is entirely up to human ingenuity, just like the selection of parameters for H.264.

> No. Patented solution is “a machine” which takes raw input from the image sensor (well, not 100% raw: it's somewhat preprocessed), stores it in some more compact form on the piece of “spinning rust” (or some other similar medium) and then plays it back.

Ah, then something which doesn't include a physical camera can't infringe the patent? I'm sure the x264 folks will be glad to know that. (You do know H.264 can be used effectively on rendered video which never touched a CCD, right?) MPEG LA claims that you need a license from them just to encode or decode H.264 video. That doesn't include capturing the raw video or presenting the decoded video on a screen. Even if it did, they didn't invent image sensors, storage devices, or displays; such things are immaterial to the patent. That would be like distinguishing between a patent on evaluating a mathematical formula and a patent on evaluation using a pencil and paper--the essence of the "something non-novel and obvious, but ON A COMPUTER!!!!" style of patent. The addition of irrelevant details should not render a patent valid when it wouldn't be valid without them.

>> As I said before, a software patent is a patent which can be infringed by writing, distributing, or executing software.
> Which nowadays includes pretty much any “traditional” patent (or perhaps you've not heard about 3D printing)?

No, in that case it's not the software that's infringing the patent. You're infringing the patent by implementing a patented physical process or producing a patented physical product. You could replace the computer with a black box, or even a human operator, and the patent would still be infringed. On the other hand, executing the same software in a simulation would not infringe any "traditional" patent.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 31, 2013 19:57 UTC (Thu) by khim (subscriber, #9252) [Link] (2 responses)

Ah, then something which doesn't include a physical camera can't infringe the patent?

It's the other way around: something which can not be used to process videos from actual physical camera and can not be used to show videos to real physical humans will not infringe.

MPEG LA claims that you need a license from them just to encode or decode H.264 video.

Well, sure. This is what the patent is all about. Think about that very-very first US patent. It explained how to better make potash and involved no machine and no apparatus! Quite literally: you take this, do that and in the end you have your pot ash and/or pearl ash! That's patent number ɛks one, the first ever accepted! Will court buy the explanation that you are not mixing ingredients by hand but use specially-built apparatus as valid? Absolutely not: anyone who sells such apparatus will need a license.

That doesn't include capturing the raw video or presenting the decoded video on a screen. Even if it did, they didn't invent image sensors, storage devices, or displays; such things are immaterial to the patent.

Well, sure. As long as image is similar to what you can receive from real-world camera and eye of viewer is similar to human eye the process will work. But if you'll feed the random noise or will try to give the result to alien with different eyes... it'll fall apart.

That would be like distinguishing between a patent on evaluating a mathematical formula and a patent on evaluation using a pencil and paper--the essence of the "something non-novel and obvious, but ON A COMPUTER!!!!" style of patent.

Sure. Now, where can I read description of techniques used by H.264 when they were used by a pencil and paper and not by computer? This will be valuable prior art.

You are right, of course, when you say that given sufficient time, you could implement H.264 in your head, with perhaps a pencil and some paper to supplement your memory. Thats absolutely true and that's the whole point: exactly because you can do that in your head the patent is valid. All these machines and devices are just “helpers” for said process. But said process itself is designed to process real-world videos for later consumption by real-world viewers—and this is why these procedures were invented in first place.

Think steam engine again: sure, blueprints probably don't infringe on a steam engine patent, but what about “ornamental details which look similarly to pieces of described patent”? If these pieces can not be used to actually build the functioning steam engine then you probably can get away with that (inflatable models don't infringe on any patents), but if you'll try to sell pieces which can actually be used to build a steam engine then court will probably not be amused.

Intent is incredibly important in court and all these talks about “software is just a math” miss that point. Because for that argument to work all the x264 developers should be able to come to court and swear that “we only did that interesting math manipulations to play with numbers, these things are not created to process real world video, this was not our intent, nope, not at all, the fact that you can actually use them to encode cat videos is just a coincidence”. This is absolutely absurd claim.

The selection of this formula rather than one of myriad alternatives which fit the facts equally well

Really? Myriad alternatives? That's interesting. What other formula of similar complexity can give as similarly usable results? Because, you see, “myriad alternatives” is exactly what patentability is all about: if there are “myriad alternatives” then patent is valid, but if there are just one (and others are either more complex or work much worse), then it's law of nature.

is entirely up to human ingenuity, just like the selection of parameters for H.264.

Really? News to me. Theory or relativity may be more precise but under certain simplifications it goes back to F=m*a which makes it basically the same thing which in turn implies unpatentability.

There is no compression algorithm, lossy or lossless, which is effective (output smaller than input) for all inputs.

Sure, but LZW patent does not explain where and how you need to dig to find that elusive “compressible data”. It does not talk about text, or video or images. It just says: here is what you are supposed to do and if you are lucky you'll achieve compression (if you are not lucky then you'll actually increase size of your data is implied). That's not something one consider as tie to real world: to say that algorithm works on data which can be successfully processed by said algorithm is tautology, it adds no useful ties to read world.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 31, 2013 23:18 UTC (Thu) by nybble41 (subscriber, #55106) [Link] (1 responses)

> Think about that very-very first US patent. It explained how to better make potash and involved no machine and no apparatus!

But it was a physical process which transformed physical matter from one state to another (and the physical change was the whole point, not just one of many ways to represent an abstract state). That is what distinguishes it from an abstract process like video encoding.

> You are right, of course, when you say that given sufficient time, you could implement H.264 in your head, with perhaps a pencil and some paper to supplement your memory.

I'm glad we agree on something.

> ... exactly because you can do that in your head the patent is valid. All these machines and devices are just “helpers” for said process.

Speeding up calculation like this is why _computers_ were invented. The patent isn't about an innovative new design for a computer, it's about an algorithm which is only useful if you happen to have a computer to speed it up. That's a big difference. Plenty of mathematical proofs can't be verified (practically) without a computer, that doesn't make them any less pure math.

> ... said process itself is designed to process real-world videos for later consumption by real-world viewers—and this is why these procedures were invented in first place.

Math with practical applications is still math.

> Think steam engine again: sure, blueprints probably don't infringe on a steam engine patent, but what about “ornamental details which look similarly to pieces of described patent”?

Again, irrelevant. You're still talking about a building a physical steam engine. Even when running the software there is no physical transformation going on which isn't part of the normal operation of the computer it's running on. Nothing novel or non-obvious, in other words.

> What other formula of similar complexity can give as similarly usable results?

Well, if you want "similar complexity", obviously there aren't very many. Most of the alternatives are much more complex. Simple doesn't necessarily mean correct, of course. You could make it simpler (e.g. F=C*a), if you were willing to accept a larger margin of error and/or a more limited domain, which is a difference of degree rather than kind. And of course there are physical laws with much more complex relationships, like the formula for the double-pendulum model, or the three-body problem. The relative ability of our eyes to resolve different visual signals would be another example.

> Sure, but LZW patent does not explain where and how you need to dig to find that elusive “compressible data”. It does not talk about text, or video or images.

So you're saying that if someone did a survey and listed which kinds of data LZW was most effective on, that would change its patentability? That makes no sense. The only reason anyone cares about LZW is that it happens to be effective for certain kinds of real-world data which people need to compress, just like H.264 is for certain kinds of video. It's relatively unspecialized in comparison to H.264, but that doesn't change its fundamental nature. There was no need to list text, video, or images because the applications were obvious. (Note that most image and video compression schemes are mainly based on something like LZW, after transformation and filtering to make the patterns in the input more apparent--that's where all the actual compression takes place.)

Cisco to release an PSEUDO-open-source H.264 codec

Posted Nov 1, 2013 0:47 UTC (Fri) by khim (subscriber, #9252) [Link]

The patent isn't about an innovative new design for a computer, it's about an algorithm which is only useful if you happen to have a computer to speed it up.

No, it's about ways to store video on a physical medium. Without compression hour of full HD video will take over terabyte of raw data. Not only it's impractical to store such file on contemporary media, it's impossible to actually watch said video because few devices can systain speeds of over 350MB/sec.

Plenty of mathematical proofs can't be verified (practically) without a computer, that doesn't make them any less pure math.

But you can verify them with a computer. Just what exactly are you are verifying with H.264?

So you're saying that if someone did a survey and listed which kinds of data LZW was most effective on, that would change its patentability?

Yes and no. If it will be just a survey it'll not change anything WRT patentability of LZW. But if you'll add such list to patent itself then yes, of course. Because such list will automatically limit the patent!

The only reason anyone cares about LZW is that it happens to be effective for certain kinds of real-world data which people need to compress, just like H.264 is for certain kinds of video.

Sure, but if you'll recall history you'll recall how the whole hoopla around LZW started: it was basically developed to transmit texts over the slow serial links, to reduce size of libraries, etc. But later it was independently created and reused for images (well, certain types of images). In a world where list of physical objects which are covered by a compression patent is limited there will be no conflict. In our world there was a conflict.

There was no need to list text, video, or images because the applications were obvious.

If it's not obvious then why not create such a list? The answer is simple: they wanted to cover the whole range of possible uses for said algorithm—and this is exactly what patents are not supposed to do. They are supposed to cover concrete and pretty limited inventions, not mathematical facts or laws of nature.

Note that most image and video compression schemes are mainly based on something like LZW, after transformation and filtering to make the patterns in the input more apparent--that's where all the actual compression takes place.

Right. But this part is “pure math” (as you like to assert) which is unpatentable. Interesting, patentable, bits are in transformations and filtering—and these things are tied to the purpose of real, tangible task of storing large amount of videodata on a small media pretty tightly. You can not use them to compress text files or audio recordings. Requirements for MPEG come from real world (MPEG targeted small CIF frames and 1.5 Mbit/s, MPEG2 targeted NTSC and PAL at 10Mbit/s, MPEG4 targets were varied, but also taken from real world), not from some abstract math ideas.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 31, 2013 0:40 UTC (Thu) by lambda (subscriber, #40735) [Link]

First of all, how many of those patents are for pure software, and how many of them actually involve some kind of specialized hardware? The ones that involve any kind of specialized hardware, such as a particular circuit for doing one of the common operations needed in encoding or decoding H.264, wouldn't count as software patents.

Furthermore, just because a patent is issued does not mean it is valid or will hold up in court when applied to a particular implementation. The mere presence of those patents does not indicate whether they would actually hold up in court.

I was intentionally vague when I mentioned "countries which don't recognize software patents." At least New Zealand falls into this category; they recently passed legislation explicitly banning software patents.

I am not an expert on European Law (especially given that there are both EU wide laws and laws and precedent in each country), but I was under the impression that it was still not entirely certain that software patents would be upheld in the EU. It's not entirely certain that they wouldn't either, and some have been issued, but I was under the impression that pure software was a lot more difficult to patent in Europe than the US.

Now, of course, there is the standard dodge of "it's not a pure software patent, it's software running on a machine." Outside of the US, how many cases do you know of in which pure distribution of software have been successfully litigated, vs. shipping hardware with that software running on it? I'm a little less concerned by the latter case, because anyone manufacturing hardware already has per-unit costs and already needs to deal with any other applicable patents on the hardware itself.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 30, 2013 14:35 UTC (Wed) by lambda (subscriber, #40735) [Link] (26 responses)

By the way, you may want to read Monty's take on the issue:

Let's state the obvious with respect to VP8 vs H.264: We lost, and we're admitting defeat. Cisco is providing a path for orderly retreat that leaves supporters of an open web in a strong enough position to face the next battle, so we're taking it.
By endorsing Cisco's plan, there's no getting around the fact that we've caved on our principles. That said, principles can't replace being in a practical position to make a difference in the future. With Cisco making H.264 available at no cost, holding out against H.264 in WebRTC makes even less sense than holding out after Google shipped H.264 in the video tag. At least under these terms, H.264 will be available at no cost to Mozilla and to any other piece of software that uses the downloadable plugin.

So yeah, they realize that this is caving on their principles, but that as a strategic move it's best to acknowledge that unencumbered codecs have lost this round of the fight, and it's best to focus their efforts on the next round.

I don't support Mozilla's decision, but I empathise

Posted Oct 30, 2013 15:27 UTC (Wed) by coriordan (guest, #7544) [Link] (23 responses)

Monty's analyses are consistently insightful.

My one-line summary of his post would be: by accepting this blob, they avoid everyone ditching Mozilla for lack of WebRTC and video, thus maintaining Mozilla's position as a conduit for a free codec to take over when h264 shows its age.

The danger, obviously, is that this form of "free software" takes off and nervous companies/consortiums might start doing this wholesale: releasing patent-shielded binaries of entire package suites like LibreOffice or GNOME or X.org, but scaring everyone away from modifying them. Distros would be reduced to arranging the binaries and distributing that arrangement. It could still be free software, but there wouldn't be that critical mass of developers needed to take control of the packages.

For packages that are under copyleft licences, we'd still have access to the real source code so we could see if there are backdoors or spy features. But in reality, if modification wasn't allowed, no one would bother to read the source code, so the whole trust/oversight system breaks down.

For non-copyleft packages (like Cisco's H264 library), there's no obligation to publish the actual corresponding source code. It's an act of faith.

I don't support Mozilla's decision, but I empathise with their position. I'll boycott Cisco's binary. No problem. But I know that many of my friends would simply move to another webbrowser, and that doesn't help the cause of free software at all.

Some clearer implications are:
* We really have to abolish software patents. Really.
* We need to fight harder for open standards (like VP8), so that Mozilla doesn't get left in situations like this.
* And, as always, we need to continue to educate about software freedom, so that users demand it. On this front, Mozilla (like most free software applications) could help by making it easier for uninformed users to discover this aspect.

I don't support Mozilla's decision, but I empathise

Posted Oct 30, 2013 15:42 UTC (Wed) by rsidd (subscriber, #2582) [Link] (7 responses)

Er, it's a BSD licence. Modification IS allowed, and redistribution, and redistribution of modifications. It satisfies all the requirements of free software. It just means you may be violating one particular patent that Cisco has licensed for their binary module. The software patent situation is such that every major piece of software probably violates a hundred patents already. If that bothers you, you shouldn't be using any software.

I don't support Mozilla's decision, but I empathise

Posted Oct 30, 2013 15:51 UTC (Wed) by rsidd (subscriber, #2582) [Link] (1 responses)

Sorry coriordan, I realise your point is a bit different and you probably agree with me on the above. Read your comment too fast (plus confused it with another - perils of mobile browsing.)

I don't support Mozilla's decision, but I empathise

Posted Oct 30, 2013 16:06 UTC (Wed) by coriordan (guest, #7544) [Link]

Ah, then my reply below doesn't apply :-)

I don't support Mozilla's decision, but I empathise

Posted Oct 30, 2013 16:05 UTC (Wed) by coriordan (guest, #7544) [Link] (4 responses)

The question isn't whether or not this is free software. The question is whether this will lead to the binary distributions of GNU/Linux distros containing packages that no one dares to modify, for fear of moving the package out from under the umbrella which protects it from patent attacks.

The reality of how software patents work is that they are only a problem if they have effects. If people are getting sued for violating them, they are a problem. If people are avoiding modifying free software because they're afraid of losing protection, then they're a problem.

Software patents currently block certain distros from distributing certain packages, and certain companies from funding development of certain packages, but, one way or another, most GNU/Linux systems end up having a working, free software H264 library which is maintained by user-developers.

Until now the Mozilla source code was 100% under the control of the Mozilla developers. Now there's 1% that comes as a binary, gets slotted into the 99%. If this becomes a trend, Mozilla could become a project that just slots together the various binary thingies it receives from various companies, and distributes that (plus source code if anyone cares to read it).

Might it be free software? Maybe. Would it be what we want? Nope.

Forks could be launched, but each would start out small and could be extinguished by a letter from MPEGLA. Divided, we're not so strong.

I assume you can see that there is at least a potential for danger here.

I don't support Mozilla's decision, but I empathise

Posted Oct 31, 2013 1:25 UTC (Thu) by JanC_ (guest, #34940) [Link] (3 responses)

Mozilla will not distribute this binary, but download it from Cisco's servers when needed (and provided you didn't disable this functionality).

Also, many distros & individuals have been distributing h.264 & other patent encumbered codecs in the past without getting into legal trouble, so I doubt they will stop doing so now. ;-)

I don't support Mozilla's decision, but I empathise

Posted Oct 31, 2013 11:59 UTC (Thu) by coriordan (guest, #7544) [Link] (2 responses)

> Mozilla will not distribute this binary, but download it from Cisco's servers

Same result.

> many distros & individuals have been distributing h.264 & other patent encumbered codecs in the past without getting into legal trouble

Yes, this is good. It undermines the patent threat by proving that it's not something people should really be afraid of, and MPEGLA might be less likely to attack such a large group.

I hope Mozilla's acts don't cause any distros to stop distributing those codecs. If the number of "unauthorised" distributors gets smaller, MPEGLA might decide to start attacking.

I don't support Mozilla's decision, but I empathise

Posted Oct 31, 2013 12:35 UTC (Thu) by khim (subscriber, #9252) [Link] (1 responses)

It undermines the patent threat by proving that it's not something people should really be afraid of, and MPEGLA might be less likely to attack such a large group.

It does nothing of the sort. It just shows that linux distributions are not large enough to warrant court expenses, nothing more, nothing less. Patents are not like trademarks, you can demand royalties for past transgression at any time.

If the number of "unauthorised" distributors gets smaller, MPEGLA might decide to start attacking.

Why? That's not how patent holders operate, you know. Some trolls may send C&D letters to mom-&-pop shops, but most will wait till you'll be large enough before suing you to make sure the whole endeavor will be profitable. This is doubly true for MPEG LA where per-seat license is small and free license is offered explicitly to the “small guys”.

long periods of non-assertion can errode patent rights

Posted Oct 31, 2013 12:55 UTC (Thu) by coriordan (guest, #7544) [Link]

> Patents are not like trademarks

If you mean that long periods of non-assertion of patents doesn't erode the rights of the patent holder, it's not as absolute as that:

http://en.swpat.org/wiki/Equitable_defences:_estoppel_and...

> Some trolls may send C&D letters to mom-&-pop shops

That's how commercial trolling works. Sending a C&D to Debian would be tactical rather than commercial since they'd know that Debian would never pay. If they taken on a whole community, they risk backlash which can take many forms. For example, many of the companies behind MPEGLA are involved in or risk being involved in anti-trust suits or investigation by the competition authorities in various countries. Since attacking Debian can't be commercial, it could be seen as eliminating a competitor. Some companies don't need that right now.

I don't support Mozilla's decision, but I empathise

Posted Oct 31, 2013 16:30 UTC (Thu) by b7j0c (guest, #27559) [Link] (14 responses)

yet monty appears to actually believe that mozilla is genuine in supporting daala and a future for free codecs. why? it will never be "practical" to use daala. mozilla is now about being "practical". almost certainly they will look to extend h.265 into firefox by an equally unpalatable arrangement.

or maybe monty realizes this and just wants to cash mozilla's checks for a few years while pursuing his hobby, which is totally understandable.

and why bother shipping firefox with vp8 etc at all at this point? why? by eich's own assertion, its a dead codec.

i'm still scratching my head as to how brendan eich was given permission to change mozilla's mission. open source was never easier to use and never about market share. i might as well use chrome.

I don't support Mozilla's decision, but I empathise

Posted Oct 31, 2013 17:03 UTC (Thu) by khim (subscriber, #9252) [Link] (13 responses)

mozilla is now about being "practical".

It was always about being “practical”. Read the original mission statement from 1998. Where does it talk about free software (or open source?) purity, about “liberty or death”, anything like that? On the contrary, it states in no uncertain terms that we realize that if we are not perceived as providing a useful service, we will become irrelevant, and someone else will take our place.

I understand your frustration: you want to send Mozilla on the wild quest and burn all it's achievements in the vain hope of creating “perfect FOSS project”, but Mozilla was never about purity. Heck, it acted as a free twin of a closed-source product from the day one! And when Mozilla Foundation was created it, again, was about creation of a successful open source project, not about cration of "a very pure free software browser". There were few tries to create "pure" free software browsers. They are either dead or irrelevant. Why don't you go and play with them instead?

i'm still scratching my head as to how brendan eich was given permission to change mozilla's mission.

This is stupid question because mission never actually changed.

open source was never easier to use and never about market share.

Again: you are trying to twist the story and put your words into another peoples mouth. Free software was never about "ease of use" or "market share". On the contrary, open source software was all about practical benefits of freely available source code with explicit goal to bring major software businesses and other high-tech industries into open source. IOW: free software was all about freedom while open source was all about practical benefits.

Today's move just highlights the fact which I've stated just recently: free software have failed while open source have won. Indeed, today's move is perfectly aligned with open-source iedology which states that open source is superior development methodology but refuses to damn closed source and, indeed, supports it when it's practical. E.g. where GCC refused to support plugins till it's hand was pushed by LLVM Firefox always supported binary plugins with [relatively] well-defined ABI

I don't support Mozilla's decision, but I empathise

Posted Oct 31, 2013 19:20 UTC (Thu) by b7j0c (guest, #27559) [Link] (6 responses)

http://www.mozilla.org/en-US/about/manifesto/

RIH

I don't support Mozilla's decision, but I empathise

Posted Oct 31, 2013 20:15 UTC (Thu) by khim (subscriber, #9252) [Link] (5 responses)

Indeed, Mozilla's manifesto now mentions FOSS (like it always did). Specifically:

7. Free and open source software promotes the development of the Internet as a public resource.

And indeed, if you'll view this one sentence in isolation then you may conclude that Mozilla have “changed their core principles”, “have sold out”, etc.

But lookie here: there is small character there! A number. Specifically number seven. Which means that there are six principles which are more important than this seventh principle! And even in said seventh principle it does not say “free software is the only way to go”. Instead it says that it's “promotes the development of the Internet”. “Development if the Internet” is clearly the goal, FOSS is just a means for that goal.

So, again, I fail to see your point. Principles number three (“The Internet should enrich the lives of individual human beings”) and six (“The effectiveness of the Internet as a public resource depends upon interoperability (protocols, data formats, content), innovation and decentralized participation worldwide”) are clearly more important in this particular case and principle number one (“The Internet is an integral part of modern life—a key component in education, communication, collaboration, business, entertainment and society as a whole”) trumps them all: if the only way to keep Internet relevant is to compromise on one of the least important principles then it must be done.

that manifesto isn't the core of this social contract

Posted Nov 3, 2013 2:30 UTC (Sun) by coriordan (guest, #7544) [Link] (4 responses)

You're correct that the manifesto is surprisingly weak in its support of free software, but I don't think anyone relies on that document for much.

People's expectations are more likely to be based on Mozilla's position in the free software community, and on the amount of support, help, and good will they've gotten from the free software community and free software projects.

Written documents are great for court cases but not necessarily for relationships. If someone feels betrayed, there's little point in holding up a document and saying: no, we've done nothing wrong, you are mistaken, we never promised to support free software!

that manifesto isn't the core of this social contract

Posted Nov 4, 2013 19:48 UTC (Mon) by gerv (guest, #3376) [Link] (3 responses)

"You're correct that the manifesto is surprisingly weak in its support of free software, but I don't think anyone relies on that document for much."

Within Mozilla, particularly among the volunteers, we rely on it rather a lot, actually. We're currently making a few judicious tweaks to push it to a 1.0 (like adding an explicit reference to privacy). The sentence about free and open source software remains the same.

that manifesto isn't the core of this social contract

Posted Nov 5, 2013 4:39 UTC (Tue) by coriordan (guest, #7544) [Link] (2 responses)

> we rely on it rather a lot, actually.

Can you give an example of how?

They're so vague. 90% of that text could easily be reused by Microsoft or Apple to justify their campaigns.

(link for others: https://www.mozilla.org/en-US/about/manifesto/ )

The only way I can imagine anyone using that text is by having their own interpretation, which would be based on Mozilla's actions. For example, the manifesto's text actually gives no firm support or commitment to free software, but based on Mozilla's actions, a volunteer could feel that principle 7 means Mozilla is committed to free software.

that manifesto isn't the core of this social contract

Posted Nov 5, 2013 11:01 UTC (Tue) by gerv (guest, #3376) [Link] (1 responses)

It's often referenced in debates about what Mozilla should do about a particular controversial topic, and it informs more specific documents such as the Privacy Principles.

I don't think Apple and Microsoft would support the Manifesto. Some examples: principle 2 implies no closed app stores. Principle 4 (once it's beefed up to specifically mention privacy) implies not tracking users secretly in order to sell them stuff. Principle 5 implies no locked devices. Principle 6 means open standards in every possible circumstance. Principle 7 says that open source is better for the Internet. Principle 8 means that software should be developed in collaboration with a community.

I'm not sure MS or Apple would sign up to any of those ideas.

that manifesto isn't the core of this social contract

Posted Nov 5, 2013 12:46 UTC (Tue) by coriordan (guest, #7544) [Link]

> I don't think Apple and Microsoft would support the Manifesto.

The principals are so soft and slippery that any good spokesperson could explain how MS or Apple helps advance them.

Of course, that wouldn't make them as nice as Mozilla. It wouldn't change anything at all.

If members of the Mozilla community manage to use that document as a tool to win arguments against non-free software, then great. But it's surely because of consensus within the community, not because of the words in that document.

For example:

"2. The Internet is a global public resource that must remain open and accessible."

Free software users could read this and hope it means rejecting non-free codecs, because they're not open (not accessible to users of free software).

MS or Apple could sign up and say it requires accepting non-free codecs because everyone should have physical/technical access to video chat.

Mozilla, it now seems, has decided it can mean either, depending on the circumstance. The words mean little or nothing.

I don't support Mozilla's decision, but I empathise

Posted Oct 31, 2013 19:23 UTC (Thu) by roc (subscriber, #30627) [Link] (5 responses)

Almost everything you wrote here is correct. The one correction I'd make is that Mozilla does care very much about software freedom. It's not our only priority, but it is very important.

The reality is that software freedom lost an important battle a long time ago when the Web (not just Web sites, but also clients with video decoding hardware) became dependent on a patent-encumbered video codec, and that wasn't Mozilla's fault. VP8 was an attempt to resolve that problem, but it hasn't worked out. Daala is our next attempt.

Whoever thinks Mozilla is "evil" for conceding ground here, ask yourself who is doing more to resolve problem of patent-encumbered video codecs so that we can return to a fully free software stack. The answer is "no-one".

For whoever really cares about this problem, and is technical (and who on LWN isn't?), Daala is a unique opportunity to attack the codec patent problem via technical contribution. You can help us improve the codec; even if you don't know much about codecs, there are a lot of ways you can help. Every bit of improvement gives Daala, and software freedom, a better chance to prevail. http://xiph.org/daala/

I don't support Mozilla's decision, but I empathise

Posted Nov 1, 2013 18:02 UTC (Fri) by mathstuf (subscriber, #69389) [Link] (4 responses)

> there are a lot of ways you can help

I understand that going onto IRC or the ML is probably the intended on-ramp here, but this is a high bar for some, so a (non-exhaustive) list on the homepage would be helpful (or a "[How to] Contribute" link on the side with a list).

I don't support Mozilla's decision, but I empathise

Posted Nov 3, 2013 16:51 UTC (Sun) by rillian (subscriber, #11344) [Link] (3 responses)

Can you be more specific about what you'd like to see? I think of IRC and mailing list as the usual way to join an open source project.

The 'Technology demos' links have general-audience level descriptions of the project appropriate for evangelists or just getting a handle on the general strategy. For a todo list, see the roadmap on the wiki.

I don't support Mozilla's decision, but I empathise

Posted Nov 3, 2013 21:08 UTC (Sun) by mathstuf (subscriber, #69389) [Link] (2 responses)

They are, but some find just jumping in to a new community to be a little anxiety-inducing (I'm getting better, but I still like to lurk a bit to test the waters).

A link directly to the roadmap would do, I think. Right now it's only accessible through Wiki -> Daala -> Roadmap (or "daala notes" -> Roadmap, but this isn't obvious IMO).

I don't support Mozilla's decision, but I empathise

Posted Nov 6, 2013 0:00 UTC (Wed) by rillian (subscriber, #11344) [Link] (1 responses)

I've added a direct link to the roadmap. Would that have been obvious enough for you to find it?

I don't support Mozilla's decision, but I empathise

Posted Nov 14, 2013 17:46 UTC (Thu) by mathstuf (subscriber, #69389) [Link]

Yep, thanks.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 30, 2013 16:10 UTC (Wed) by johnsmith_notthatone (guest, #92971) [Link] (1 responses)

Thank you for your comment. Monty article is honest and informative. It doesn't evade responsibility or whitewashes the situation.

And it includes two other pieces of important information:

«That said, today's arrangement is at best a stopgap, and it doesn't change much on the ground. How many people don't already have H.264 codecs on their machines, legit or otherwise? Enthusiasts and professionals alike have long paid little attention to licensing. Even most businesses today don't know and don't care if the codecs they use are properly licensed[1]. The entire codec market has been operating under a kind of 'Don't Ask, Don't Tell' policy for the past 15 years and I doubt the MPEG LA minds. It's helped H.264 become ubiquitous, and the LA can still enforce the brass tacks of the license when it's to their competitive advantage (or rather, anti-competitive advantage; they're a legally protected monopoly after all).»

«The giveaway also solves nothing long-term. H.264 is already considered 'on the way out' by MPEG, and today's announcement doesn't address any licensing issues surrounding the next generation of video codecs. We've merely kicked the can down the road and set a dangerous precedent for next time around. And there will be a next time around.»

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 30, 2013 19:25 UTC (Wed) by mathstuf (subscriber, #69389) [Link]

I think you missed a reference for your "[1]".

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 30, 2013 14:54 UTC (Wed) by rsidd (subscriber, #2582) [Link] (11 responses)

We know who Mozilla is and what they have done for free software. Who are you?

useful subject line

Posted Oct 30, 2013 15:33 UTC (Wed) by coriordan (guest, #7544) [Link] (1 responses)

Do you need his identity to be able to decide if he's right or wrong?

Mozilla has done lots for free software, but that doesn't mean we can't analyse individual decisions. And if a number of individual decisions go in the wrong direction, I see nothing wrong with pointing this out.

useful subject line

Posted Oct 30, 2013 15:45 UTC (Wed) by rsidd (subscriber, #2582) [Link]

The GP was flamebait. Your other comment above was a bit better and I replied there.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 30, 2013 16:19 UTC (Wed) by johnsmith_notthatone (guest, #92971) [Link] (8 responses)

I have a conscience. Do you?

They stated their principles, they said they would defend them. Now they betrayed them.

I didn't forced their hands one way or another.

I've been using Mozilla from the very beginning, the Netscape open source project. At that time it was barely functioning. I even contributed a little, very little, translations.

I stood by them during the difficult periods because I supported their principles, now they are moving away from them. Should I stay quiet?

Your reply doesn't do anything to improve the discussion. It's not civil discourse.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 30, 2013 16:42 UTC (Wed) by lambda (subscriber, #40735) [Link] (6 responses)

This is really no different than shipping a plugin API that allows binary plugins. Right now you can download a licensed H.264 decoder in the form of Flash. In the future, you'll be able to do so in the form of Cisco's H.264 decoder which can work with the HTML5 APIs, and will have source code available under the MIT license. This seems like a pretty big win over the Flash solution; it's obviously not ideal, it means they lost the battle for an open codec as the standard, but it's an improvement over the status quo which is using Flash for this kind of functionality.

Better for them to spend their efforts on the Daala front than to continue to fight this battle. If Daala actually delivers what it promises, a real next-gen codec that surpasses H.265, and is available royalty free and stable within a year or so, it could have a serious chance of displacing H.265. My main worry is that it may arrive too late; by the time it's stable, H.265 hardware development is likely to have been in the works for a couple of years. VP9 is available now, approximately on par with H.265, so it may be a stopgap that could be used in the meantime, it's hard to say. It can just be a problem to have to switch standards so often; it's too bad Daala is not yet ready.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 30, 2013 16:53 UTC (Wed) by johnsmith_notthatone (guest, #92971) [Link] (5 responses)

I'm afraid that Daala is still too far into the future. VP8 and VP9 are available now and they very little chance because of the recent support for H264. H265 is the natural continuation, both from the hardware and software fronts. It'll win by inertia. AAC didn't displaced MP3 for the same reason, despite bigger support.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 31, 2013 16:35 UTC (Thu) by b7j0c (guest, #27559) [Link] (4 responses)

daala died yesterday too. deep down i am sure monty realizes this. mozilla is telling us they see no value in taking a risk on a minority codec for the sake of freedom. so how would daala ever get off the ground? every "network effects" argument against vp8/9 will be made against daala. everyone else who might care is already a h.264 license holder (or better, an mpeg-la member) who will just migrate to the known-quantity of h.265.

working on daala still might be fun and educational, but no longer of any value to users

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 31, 2013 18:17 UTC (Thu) by mathstuf (subscriber, #69389) [Link]

Hell, I thought *I* was a pessimist. VP8 wasn't (technologically) better than H.264 (or at least enough) to overcome those network effects. Daala has the potential to be much better than H.265 and VP9 since it's not just incremental improvements on old algorithms while at the same time avoiding the H.264 patent morass because they're completely different algorithms. Opus has succeeded in pretty much bettering codecs from low to high bitrates and get wide adoption, why can't Daala do similarly if it is shown to be better in similar ways?

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 31, 2013 18:57 UTC (Thu) by dlang (guest, #313) [Link]

given that they aren't trying to compete against VP8/VP9, but rather against VP10 and that generation, there is no entrenched competition that they are so far behind, rather they are on a level playing field at the moment, and per this report, things are looking good.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 31, 2013 19:49 UTC (Thu) by roc (subscriber, #30627) [Link] (1 responses)

Mozilla "took a risk on a minority codec for the sake of freedom" for YEARS, while no-one else did. (Google infamously said they would, by removing H.264 from Chrome, but quietly decided not to.)

It is clear that Mozilla wasn't powerful to change the codec landscape just by saying "we're not going to support H.264 natively because it's patent encumbered (though we will support it via Flash)" and standing alone on that. We tried that strategy for years and it had no effect. With Daala we're trying to do things differently:
-- create a codec that performs better than the competition
-- create a codec that uses quite a different design and is hopefully less attackable via patents
-- create a codec that isn't controlled by a single vendor who is hated and feared by the rest of the industry
-- create a codec that is properly standardized (through the IETF probably)
-- use a truly open development model
-- build on what we learned (both technical and political) with Opus, which has been very successful

The environment for Daala vs HEVC/H.265 will be somewhat different than we faced with VP8 vs H.264. H.264 will already be available as a good baseline so declining to support HEVC (on the server or client) is a much more reasonable proposition than declining to support H.264 was. HEVC hardware and licensing aren't available yet so we're not as far behind as we were with H.264/VP8. And who knows, maybe this time Google will stick to their guns.

Sure, fighting the video patent cartel is an uphill battle and we might not win. But we're winning with Opus and we think we can win again with Daala. And this time, people can get involved and help Daala win instead of just carping from the sidelines. https://xiph.org/daala/

Cisco to release an PSEUDO-open-source H.264 codec

Posted Nov 2, 2013 14:57 UTC (Sat) by basilgohar (guest, #93725) [Link]

I agree wholeheartedly with Monty and Mozilla in general that Daala is the right way to fight this fight for the future, for all the reasons listed above and more.

However, I think the fight for VP8 as the MTI codec in WebRTC is not yet over, as the Cisco offer has sparked a very large number of people to share their concerns with this less-than-perfect solution, and, in fact, their preference for VP8 as the codec of choice to be MTI:

http://www.ietf.org/mail-archive/web/rtcweb/current/msg09...

That entire thread, in fact, is filled with many useful points criticising the Cisco offer in light of the intended goals of an MTI codec in what is supposed to be an open Internet standard.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Nov 4, 2013 8:22 UTC (Mon) by jezuch (subscriber, #52988) [Link]

It's one thing to stand by one's principles; it's another to be hard-headed about it and deny the reality. Sometimes you DO have to retreat and regroup.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 30, 2013 15:09 UTC (Wed) by Otus (subscriber, #67685) [Link]

> Because the audio part is missing:

What does this mean in practice... Should we now use H.264 for video and Vorbis for audio? Opus for audio? What's the best choice if you care about supporting both free software systems and popular mobile devices?

Cisco to release an PSEUDO-open-source H.264 codec

Posted Oct 30, 2013 15:25 UTC (Wed) by shmerl (guest, #65921) [Link]

Audio part is covered with Opus which is a free codec. It's a mandated part of WebRTC. Video however has no mandated codec so far. Hopefully when Daala comes out, it will be added to the standard as mandated as well.

Cisco to release an PSEUDO-open-source H.264 codec

Posted Nov 4, 2013 19:38 UTC (Mon) by gerv (guest, #3376) [Link]

It's important to realise that the Mandatory To Implement coded for WebRTC is Opus - the best audio codec out there, which is also open and free. So the "without an audio part" criticism doesn't apply to WebRTC.

It does apply to the most common ways of using H.264 video in the <video> tag; getting that working fully requires a license for the AAC codec. Nothing on that has been announced... yet.

What option would you have Mozilla take instead? Continue to deadlock the WebRTC group so that entirely proprietary methods of communication get more of a hold? Continue to refuse to support H.264 in <video> and watch all the sites which use it recommend alternative browsers, so Firefox's influence over _anything_ dissipates with its market share? The options here are not good.

Cisco to release an open-source H.264 codec

Posted Oct 30, 2013 15:34 UTC (Wed) by CopperWing (guest, #82856) [Link] (14 responses)

I think everybody is missing the most important point of the announcement: Cisco said they will pay the royalties to MPEG for all the usage of this codec done by open source software and will act as a licensing proxy, so to provide H.264 under licensing terms compatible with open source software.

Cisco to release an open-source H.264 codec

Posted Oct 30, 2013 16:06 UTC (Wed) by johnsmith_notthatone (guest, #92971) [Link] (5 responses)

The release is divided in two parts.

The first is the source code. This have the same side effects of the open source implementation X264. Any payments have to be negotiated directly with MPEG-LA.

The second is the binary. It's gratis. Downloaded directly from Cisco and can't be modified. This changes nothing to open-source.

Neither Fedora or Debian will include any of them. The first because of patents and royalties. The second because is not open source.

This exactly the same situation that we had with Silverlight/Moonlight. The codecs had to be downloaded directly from Microsoft. Everyone was up in arms back then, but now everyone is happy! It seems that Miguel de Icaza was right all along. We as a community are bunch of hypocrites.

Cisco to release an open-source H.264 codec

Posted Oct 30, 2013 16:32 UTC (Wed) by coriordan (guest, #7544) [Link] (4 responses)

I think Fedora and Debian could include this binary in their binary distribution and the non-protected source code in their source distribution, without violating their charters regarding free software, if they could prove that the source code was the exact corresponding source code. Fedora's strict patent policy would surely prevent this though. Debian's patent policy is less rigid.

But I'd argue that they shouldn't.

Moonlight was a different case because Silverlight's market share was weak (proof: it's disappeared). Promoting Moonlight implied encouraging websites to write Silverlight-compatible widgets, giving Microsoft more control over www technologies. The best way to defeat Silverlight was to ignore it. We did and we won. H264 is different because it's already (sadly) very well established. If we push too hard, we might just push ourselves backwards rather than harming h264.

De Icaza tries to convince free software users to relativise their convictions because he makes money from selling proprietary software. Trying one's best but not achieving 100% doesn't make one a hypocrite. De Icaze was worth listening to in the 90s. Not sure what changed him.

Cisco to release an open-source H.264 codec

Posted Oct 30, 2013 16:45 UTC (Wed) by johnsmith_notthatone (guest, #92971) [Link] (1 responses)

My comparison was not in relation with moonlight, which I was against for the same reasons as you, but with "downloading binary codecs from third party". The latter was what made the situation unacceptable, given that moonlight was open source.

I stand by my comparison, since they're similar.

Cisco to release an open-source H.264 codec

Posted Oct 30, 2013 17:13 UTC (Wed) by coriordan (guest, #7544) [Link]

But the free software community didn't accept the proposal to download binary codecs for Silver/Moonlight from a third party.

Cisco to release an open-source H.264 codec

Posted Oct 30, 2013 17:22 UTC (Wed) by dcbw (guest, #50562) [Link] (1 responses)

"I think Fedora and Debian could include this binary in their binary distribution and the non-protected source code in their source distribution, without violating their charters regarding free software, if they could prove that the source code was the exact corresponding source code."

No, Fedora can't. The source code that's distributed in Fedora must be used to build the distribution. You cannot distribute these pre-built binaries even if you could somehow prove they are the same, which is pretty hard.

http://fedoraproject.org/wiki/Packaging:Guidelines#No_inc...

correct. but cisco's driver will come in the back door

Posted Oct 31, 2013 13:04 UTC (Thu) by coriordan (guest, #7544) [Link]

Thanks for the correction.

There was another error in my comment. Contrary to what I implied, Mozilla won't be including this blob. Their software will instead download it when run.

Debian already includes software which installs binary blobs from the internet when you run it, such as Firefox which quickly prompts you to click to download Flash, or the hplip package which immediately prompts you to download a proprietary HP printer blob driver, so they might also turn a blind eye to their free software guidelines and accept this Cisco situation. (This is one of the reasons why FSF doesn't consider Debian to be a 100% free software distro.)

Cisco to release an open-source H.264 codec

Posted Oct 30, 2013 16:07 UTC (Wed) by lambda (subscriber, #40735) [Link] (6 responses)

Not compatible with all FLOSS. For instance, you couldn't use it in a GPLed application, since you are not able to grant your users the patent license that would be necessary to comply with the terms of the GPL about being able to rebuild and redistribute binary forms of the GPLed application.

Cisco to release an open-source H.264 codec

Posted Oct 30, 2013 19:37 UTC (Wed) by mathstuf (subscriber, #69389) [Link] (5 responses)

You couldn't use it from GPLv3, but I don't think there's a problem with GPLv2. Apache might have issues though if you already idemnify your downstreams for patent issues.

Cisco to release an open-source H.264 codec

Posted Oct 30, 2013 20:36 UTC (Wed) by lambda (subscriber, #40735) [Link] (4 responses)

From the GPLv2:

If, as a consequence of a court judgment or allegation of patent infringement or for any other reason (not limited to patent issues), conditions are imposed on you (whether by court order, agreement or otherwise) that contradict the conditions of this License, they do not excuse you from the conditions of this License. If you cannot distribute so as to satisfy simultaneously your obligations under this License and any other pertinent obligations, then as a consequence you may not distribute the Program at all. For example, if a patent license would not permit royalty-free redistribution of the Program by all those who receive copies directly or indirectly through you, then the only way you could satisfy both it and this License would be to refrain entirely from distribution of the Program.

The H.264 patent license does not permit royalty-free redistribution. Thus, if you distributed GPLv2 code with this code linked to it, you would be distributing something with a more restrictive license than the GPLv2, which is not allowed. I would be curious to see an argument against this, but from my reading, you can't link patent-encumbered code from GPLed code.

Cisco to release an open-source H.264 codec

Posted Oct 30, 2013 22:10 UTC (Wed) by khim (subscriber, #9252) [Link] (2 responses)

Yes, it's perfectly legal, but it's quite complex dance.

First, the BSD source. It does not include any patent languages at all. And you can use it freely for development.

Ask Cisco about the situation and it'll say: we don't know if MPEG LA's patents are infringed by this source code and we don't particularly care. It's probably not infringing, but we just don't know. You must talk to MPEG LA guys, we can not say if it's infringing or not. It's their patents, not ours, after all!

What's GPL-incompatible about this situation? Answer is: nothing, of course.

Ask MPEG LA about the situation and you'll hear: this code probably infringes, but we just don't know for sure. And we refuse to answer the infringement question, but as long as you are not distributing anything (or only give few copies it to your friends from time to time) you either don't infringe or fall under the cap (there are free license for “small users” and if you are not distributing anything you are small user). In both cases you are free to do what you are doing.

Next step. Binary blob. Binary which is distributed is a proprietary blob and it's license is most definitely not GPL-compatible, but you can not distribute binary of a GPLed program with this module included anyway. It must be downloaded from the Cisco's server by each user explicitly and separately. This time situation is the opposite: since binary was downloaded from Cisco's servers GPL is not involved (hey, it's Cisco's code and Cisco can distribute it's software under any license it wants to, proprietary or not). If you'll call any program which can load and use binary modules illegal will be crazy: think NDISwrapper. You can use it to load proprietary binary modules in your kernel, but it does not make you kernel illegal, right? Here it's the same situation. User is covered as well: The act of running the Program is not restricted, and the output from the Program is covered only if its contents constitute a work based on the Program (independent of having been made by running the Program).

Again: Cisco will say that said blob is “maybe infringing or maybe not (but probably not)”, but it does not matter because in both cases you are free to use it and MPEG LA will say that said blob “maybe infringing or maybe not (but probably infringing)” but in both cases you are covered (because Cisco paid “the flat rate” and now can distribute anything it wants without complains from MPEG LA).

See? Without actually going to court you can not claim that any GPL-incompatible terms are involved at any stage. But if you'll grab BSD-licensed sources and try to use binary compiled from said sources then you'll probably find out that you are not covered and must pay license fee to MPEG LA, but even that will change nothing for other players. For, you see, software is unpatentable. Only “machine” with software can infringe. Which means that your decision will only cover “machine” (mobile phone with Firefox clone installed or some kind of Linux-based VCR, etc). To see if other devices will infringe or not, you'll need another victim^H^H^H^Hvolunteer—and it, again, will cover either tablet (with some clone of Chromium, installed, naturally) or dish-washer (with some kind of VLC installed, of course), but never Chromium, Firefox, or VLC itself. Worst-case scenario: court will decide that any device with Firefox installed is infringing (but then, of course, that particular version of Firefox which is bundled with BSD-licensed codec, not version of Firefox which downloads codec from Cisco's servers).

P.S. What happens if you'll try not to distribute binaries and will only distribute sources? Well, if binary will be built by some automated script or can be produced by following some instruction then MPEG LA can, naturally, claim that you actually distribute the binary which is produced by such system and that binary is included in the “machine” in question. Said binary is just distribute in, you know, unassembled form. after all people buy furniture which requires assembly all the time, why software should be any different? And, again, they will start to care only if you'll be big enough (smaller guys are covered by “first dose is free license” anyway).

Cisco to release an open-source H.264 codec

Posted Oct 31, 2013 10:47 UTC (Thu) by etienne (guest, #25256) [Link] (1 responses)

> But if you'll grab BSD-licensed sources and try to use binary compiled from said sources then you'll probably find out that you are not covered ...

I surely do not want to be the guy trying to explain the court what is compilation, that compilation is different from early/late linking and module inserting, that any simple assembly instruction in the executable can be intercepted and changed at run time to do something completely different, and that a software compiled for a machine can be executed on a complete different machine (qemu,...). Even the source can be executed by a C interpreter or a just-in-time compiler.
I am not sure the court will tell me at which point the patents rights are no more transferred (code has been interrupted by the timer interrupt, do I still have a license?), one of their possible question is: Why does it matter?
It would obviously matter if the binary module is not the representation of the source, like some backdoor added.

Cisco to release an open-source H.264 codec

Posted Oct 31, 2013 13:04 UTC (Thu) by mathstuf (subscriber, #69389) [Link]

It can be shown that the source corresponds wish the binary. I'd be easier if we knew what the tool chain is, but I'm sure someone will do this (as was done with TrueCrypt recently).

Cisco to release an open-source H.264 codec

Posted Oct 31, 2013 13:10 UTC (Thu) by jsanders (subscriber, #69784) [Link]

> The H.264 patent license does not permit royalty-free redistribution. Thus, if you distributed GPLv2 code with this code linked to it, you would be distributing something with a more restrictive license than the GPLv2, which is not allowed. I would be curious to see an argument against this, but from my reading, you can't link patent-encumbered code from GPLed code.

I'm no lawyer, but if your GPL program supported a standard video plugin API, then I wouldn't consider the combined software + plugin to be covered by the GPL, providing that the program did not have to use the non-GPL plugin. In the same way, I'm not convinced that linking to GPL code which provides an API (e.g. readline) which is also provided by other non-GPL libraries, really makes the whole program covered by the GPL. You could also use interprocess communication to separate the GPL and non-GPL code. Does a propitiatory IDE using gcc via the shell break the GPL?

What makes you say that?

Posted Oct 30, 2013 16:36 UTC (Wed) by coriordan (guest, #7544) [Link]

I think it's not clear what's covered. What sentence in Cisco's blog post gives you such hope?

What we need, is to see that licences and legal texts. (AFAICS they're not available.)

Timing is odd

Posted Oct 30, 2013 16:12 UTC (Wed) by bawjaws (guest, #56952) [Link] (4 responses)

So, I'm generally sympathetic to Firefox's position here, but one thing sticks out to me, and that's the timing of the announcement.

There's no code, no binary, nothing but promises and plans today. It would appear the main reason for talking about it now, rather in many months when it's a shipping thing with concrete details (quality of code, quality of encoding, licence terms etc.) is to influence the decision on which codecs would be mandatory to implement in WebRTC.

The big options are "VP8", "H.264 (baseline)" & "No MTI video codec".

What does Mozilla get from being part of the ad campaign for "H.264 (baseline)" as MTI codec? It could quite easily have supported VP8 right up until the decision was made and then, regardless of the exact decision, adopted Cisco's binary plugin for interoperability. Did Cisco extract this support in exchange for their donation?

I'm hoping that "VP8" is chosen, but since IETF runs on consensus it currently seems more likely that "No MTI video codec" is the result. But I'd be a bit miffed if I felt Mozilla's involvement in this tipped things towards H.264 and meant that in 5 or 10 years time people were still paying license fees for H.264 just to be compliant when they were actually using a superior codec like VP9, Daala or even H.265.

Timing is odd

Posted Oct 30, 2013 16:27 UTC (Wed) by johnsmith_notthatone (guest, #92971) [Link] (3 responses)

«What does Mozilla get from being part of the ad campaign for "H.264 (baseline)" as MTI codec? It could quite easily have supported VP8 right up until the decision was made and then, regardless of the exact decision, adopted Cisco's binary plugin for interoperability. Did Cisco extract this support in exchange for their donation?»

I wasn't aware that a vote on the matter was imminent. It already was strange that Mozilla made such U-turn on their position. But with this background it becomes even more disturbing.

Do you have any evidence of any agreement between Cisco/Mozilla beyond your suspicion, or that the announcement implied Mozilla support for H264 position.

Timing is odd

Posted Oct 30, 2013 17:04 UTC (Wed) by bawjaws (guest, #56952) [Link] (2 responses)

The vote is in one week's time:

https://datatracker.ietf.org/meeting/88/agenda/rtcweb/

Mozilla and Cisco seem to have co-ordinated on the planning of this announcement. I see no benefit to Mozilla or the web to announce their involvement this week rather than next. I do see a benefit to those who wish H.264 to be made mandatory (or equally for VP8 to not be mandatory).

Those are the dots I'm connecting, but I have nothing else to back them up.

Timing is odd

Posted Oct 30, 2013 17:42 UTC (Wed) by raven667 (subscriber, #5198) [Link]

they are probably just preemptively reacting to something which has already happened

It seems obvious to me...

Posted Oct 30, 2013 19:20 UTC (Wed) by thams (guest, #93686) [Link]

It seems pretty obvious to me: Mozilla's cooperation must have been a condition either on the whole deal or on some detail of it like the binary being something everyone could use on all platforms and by all parties vs being licensed for Mozilla use only or availability of source-code.

Tip of an iceberg?

Posted Oct 30, 2013 16:33 UTC (Wed) by cesarb (subscriber, #6266) [Link] (5 responses)

I'm scratching my head here. It seems there is a lot more going on that most of us can see. Monty probably knows what is really happening, but cannot tell for now.

The first question would be "what is in it for Cisco?" I can see at least two possibilities. Either they already reach the yearly cap with their own proprietary products, so the cost of throwing the open source crowd a bone is next to nothing. Or, given that they are involved with Daala, they are playing some overly complicated chain of chess moves, and this is just the first of them. For example: they are doing this to strengthen H.264, in a way which weakens the proprietary H.264 successor (why use it if H.264 is good enough and Cisco's move just made it better in the "I can use it everywhere" metric?), therefore making it easier for the proprietary H.264 successor's non-proprietary competitor to win. Or something even more complicated.

Tip of an iceberg?

Posted Oct 30, 2013 16:45 UTC (Wed) by lambda (subscriber, #40735) [Link]

Cisco sells lots of audio and videoconferencing products. They probably want to be able to add WebRTC support, so that people can listen or join via the web browser. But with a fragmented codec landscape, that becomes hart. By doing this, they allow one of the most popular browsers, which has support for WebRTC, to support H.264, the same standard that lots of Cisco's hardware products already support.

Tip of an iceberg?

Posted Oct 30, 2013 17:43 UTC (Wed) by b7j0c (guest, #27559) [Link]

cisco is in the mpeg-la. when the bear-trap is inevitably sprung on the users of the web, cisco will reap the benefit along with the other patent pool members

Benefit to Cisco

Posted Oct 30, 2013 22:10 UTC (Wed) by Tester (guest, #40675) [Link]

The benefit to Cisco is simple. They sell MCU (videocall conference servers) for a lot of money, these use (are based on) H.264 and I guess they want to be able to keep on selling them with WebRTC endpoints. And we can assume that the cost is relatively small, as they are already at the licensing cap or very close. So I wouldn't be surprised if one year of sales of MCUs for WebRTC would cover their entire licensing costs.

Tip of an iceberg?

Posted Oct 31, 2013 21:41 UTC (Thu) by marcH (subscriber, #57642) [Link]

The above answers are basically correct. Cisco is on the Telepresence warpath. They hate themselves for having mostly lost the first battle to Skype/Facetime/Hangouts so now they are playing the cards typically played by minor players: interoperability, compatibility, openness,...

Exactly like how Android beat iOS. Of course only until http://lwn.net/Articles/571040/

Tip of an iceberg?

Posted Nov 4, 2013 20:26 UTC (Mon) by gerv (guest, #3376) [Link]

Cisco do not reach the cap already, so it will cost them.

Cisco could release an open-source x264 codec

Posted Oct 30, 2013 18:00 UTC (Wed) by pauliusz (guest, #49461) [Link] (9 responses)

It would be better if Cisco provided x264 codec blob.
That would be bigger win for open source.

Cisco could release an open-source x264 codec

Posted Oct 30, 2013 18:18 UTC (Wed) by khim (subscriber, #9252) [Link] (8 responses)

It's impossible because of x264 license (GPLv2).

Cisco could release an open-source x264 codec

Posted Oct 30, 2013 19:26 UTC (Wed) by gmaxwell (guest, #30048) [Link]

For some definitions of impossible. http://x264licensing.com/

Cisco could release an open-source x264 codec

Posted Oct 30, 2013 21:38 UTC (Wed) by krake (guest, #55996) [Link] (6 responses)

If providing a binary for a GPLv2 licenced program would be impossible, we would all be using Gentoo.

Cisco could release an open-source x264 codec

Posted Oct 30, 2013 23:28 UTC (Wed) by khim (subscriber, #9252) [Link] (5 responses)

It's of course possible to provide compiled binary of GPLv2 code, but you are not allowed to restrict redistribution. And binaries offered by Cisco must include such limitation (this is requirement of MPEG LA license). With BSD it's easily solvable: source is provided on terms of BSD license, but it does not include patent grant, binary (compiled from that same sources!) is distributed under different license (with redistribution forbidden) but include patent grant.

As was noted Cisco could have arranged for a separate, non-GPL license for x264, but I guess it was easier for them to release their own stuff rather then to buy license for x264. And since their goal is to make sure WebRTC is compatible with their implementation…

Cisco could release an open-source x264 codec

Posted Oct 31, 2013 7:56 UTC (Thu) by krake (guest, #55996) [Link] (3 responses)

The unrestricted redistribution applies mainly to the sources.
The binary can often be packaged in a way that the package does not allow direct redistribution as such.

My guess is that it is just easier for them to work with their own code, especially considering that they have to build it for so many platforms.

Cisco could release an open-source x264 codec

Posted Oct 31, 2013 10:12 UTC (Thu) by khim (subscriber, #9252) [Link] (2 responses)

The unrestricted redistribution applies mainly to the sources.

Really? That's quite novel interpretation of clause Each time you redistribute the Program (or any work based on the Program), the recipient automatically receives a license from the original licensor to copy, distribute or modify the Program subject to these terms and conditions. You may not impose any further restrictions on the recipients' exercise of the rights granted herein.

Redistribution rights are granted by GPL to both binaries and sources. You can combine GPLed code and non-GPLed code to produce something which can not be redistributed as a whole, that's true, but GPLed binaries themselves can always be redistributed. And MPEG LA requirements are the exact opposite: no redistribution rights are ever given, each copy must be created by the licensor (but then can be propagated via the distribution chain).

Cisco could release an open-source x264 codec

Posted Nov 2, 2013 18:04 UTC (Sat) by krake (guest, #55996) [Link] (1 responses)

> Redistribution rights are granted by GPL to both binaries and sources.

Which is why I wrote "mainly" not "only".

Because most binary distributions are not just the program but contain the program and other things, each potentially having a different license.

Cisco could release an open-source x264 codec

Posted Nov 3, 2013 18:20 UTC (Sun) by khim (subscriber, #9252) [Link]

It does not change anything for Cisco. Since Cisco needs a license which does not give you right to redistribute binaries GPL's “freedom or death” clause here most definitely means “death”. Cisco can not do what it does and still use GPL. It can probably ask x264 authors for some different license, but at this point it's probably easier to just use something else.

BSD License does not have such problem: it just says “here are the sources, do whatever you want with them”. You still need separate patent license and BSD does not forbid that.

Cisco could release an open-source x264 codec

Posted Oct 31, 2013 12:22 UTC (Thu) by oshepherd (guest, #90163) [Link]

Doing this with x264 requires that they reach an agreement to do this with the x264 developers.

Given that the x264 developers get quite a bit of money from selling licenses to use x264 under non-GPL licenses, it's not like this is an arrangement they could exactly agree with.

Besides, the BSD license makes this code more widely useful (though its somewhat irrelevant being as this blob is being used under its' own blob license, not the open source license, Mozilla, for one particularly important example, cannot depend upon GPL code)

Cisco to release an open-source H.264 codec

Posted Oct 30, 2013 21:13 UTC (Wed) by xnox (subscriber, #63320) [Link] (3 responses)

Binary decoder or encoder or both are getting released?
As far as I remember, royalties for encoding are high.

Cisco to release an open-source H.264 codec

Posted Oct 30, 2013 22:34 UTC (Wed) by khim (subscriber, #9252) [Link]

Pure decoder will be pretty useless for WebRTC, don't you think?

Cisco to release an open-source H.264 codec

Posted Oct 31, 2013 1:51 UTC (Thu) by JanC_ (guest, #34940) [Link]

According to Monty of Xiph.org, you never have to pay MPEG LA more than about 6.5mln (the license cost is capped), and Cisco is probably already paying close to that because of the video conferencing software & hardware they sell, so the cost for them to distribute this codec for free might be considerably less than it seems at first sight.

Cisco to release an open-source H.264 codec

Posted Nov 4, 2013 10:32 UTC (Mon) by DonDiego (guest, #24141) [Link]

There are codec licensing schemes where the fees for encoding are higher than for decoding, but H.264 is not one of them.

Cisco to release an open-source H.264 codec

Posted Oct 31, 2013 0:29 UTC (Thu) by heijo (guest, #88363) [Link] (2 responses)

Why are they reinventing the wheel instead of just using ffmpeg+x264?

Cisco to release an open-source H.264 codec

Posted Oct 31, 2013 5:09 UTC (Thu) by Kamilion (subscriber, #42576) [Link]

To ensure they work with Cisco videophones, of course.

And if the whole thing plays out like Asterisk and SKINNY/SCCP support, who knows.
Hopefully WebRTC will swoop in and save the day and it won't just be one more footnote in the story of "Why the video-phone took 80 years to become commonplace".

Cisco to release an open-source H.264 codec

Posted Oct 31, 2013 12:20 UTC (Thu) by oshepherd (guest, #90163) [Link]

Because FFMPEG/LibAV/x264 are (L)GPL wheels which can't be redistributed in this manner (Cisco would have to give a license grant to all downstream recipients, which their MPEG-LA license will not permit)

Cisco to release an open-source H.264 codec

Posted Oct 31, 2013 20:24 UTC (Thu) by marcH (subscriber, #57642) [Link] (1 responses)

> ... and to provide it as a binary module that can be downloaded for free from the Internet.

Hi, Fort Meade!

Cisco to release an open-source H.264 codec

Posted Oct 31, 2013 20:32 UTC (Thu) by marcH (subscriber, #57642) [Link]

Or maybe not

http://xiphmont.livejournal.com/61927.html?thread=442343#...

Cisco to release an open-source H.264 codec

Posted Nov 1, 2013 1:51 UTC (Fri) by gdt (subscriber, #6284) [Link] (1 responses)

The Cisco link is well worth a reading as a critical text. Look at the quality of the responses from this huge technology company: it clearly shows a deep understanding of the FOSS movement, its licensing, its aspirations, its woes. Consider if Cisco would have been capable of this quality of discussion even five years ago.

Consider that someone in Cisco thought of this evil plan; was capable of explaining it within Cisco; and could convince enough managers that the financial risk was worth the benefit (basically that FOSS would bring enough users for sales of videoconferencing and telepresence equipment to cover the increased payments to MPEG-LA).

It's not a great day for FOSS. But we can take some solace that after decades of being misunderstood, free software now so well and truly understood by the commercial sector that plans like this can be made, approved, and executed.

Cisco to release an open-source H.264 codec

Posted Nov 1, 2013 10:24 UTC (Fri) by cladisch (✭ supporter ✭, #50193) [Link]

> … to cover the increased payments to MPEG-LA

The license caps the payments at (currently) $6.5 million per year.

It's very likely that Cisco has already reached this limit, and that it can distribute the codec literally for free.

I predict that MPEG-LA will try to close this loophole in the licenses of future video codecs; Daala will need to compete against those.

Cisco to release an open-source H.264 codec

Posted Nov 6, 2013 11:33 UTC (Wed) by ssam (guest, #46587) [Link]

Suppose debian have a package containing the source, that can build a binary that is bit for bit the same as the binary direct from CISCO. Presumably from debians point of view that is just the same as having a package for x264, and from CISCOs point of view you are not covered because you did not get the binary from them.

Now you have a script called get_licensed_binary, that downloads the binary from CISCO, and writes it on top of the one from the debian package. That would make you covered by the licence, but could not be included in the debian package, because its installing a binary blob.

What about a script called verify_binary, that downloads CISCOs binary, and checks bit by bit that they are identical. does that cover you by the licence? what if the script would fix any bit that was different (of course there would be none).

Cisco to release an open-source H.264 codec

Cisco to release an open-source H.264 codec

Cisco to release an open-source H.264 codec

Cisco to release an open-source H.264 codec

Cisco to release an "open-source" *binary-only* H.264 codec

Cisco to release an "open-source" *binary-only* H.264 codec

Cisco to release an "open-source" *binary-only* H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

I don't support Mozilla's decision, but I empathise

I don't support Mozilla's decision, but I empathise

I don't support Mozilla's decision, but I empathise

I don't support Mozilla's decision, but I empathise

I don't support Mozilla's decision, but I empathise

I don't support Mozilla's decision, but I empathise

I don't support Mozilla's decision, but I empathise

I don't support Mozilla's decision, but I empathise

long periods of non-assertion *can* errode patent rights

I don't support Mozilla's decision, but I empathise

I don't support Mozilla's decision, but I empathise

I don't support Mozilla's decision, but I empathise

I don't support Mozilla's decision, but I empathise

that manifesto isn't the core of this social contract

that manifesto isn't the core of this social contract

that manifesto isn't the core of this social contract

that manifesto isn't the core of this social contract

that manifesto isn't the core of this social contract

I don't support Mozilla's decision, but I empathise

I don't support Mozilla's decision, but I empathise

I don't support Mozilla's decision, but I empathise

I don't support Mozilla's decision, but I empathise

I don't support Mozilla's decision, but I empathise

I don't support Mozilla's decision, but I empathise

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

useful subject line

useful subject line

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an PSEUDO-open-source H.264 codec

Cisco to release an open-source H.264 codec

Cisco to release an open-source H.264 codec

Cisco to release an open-source H.264 codec

Cisco to release an open-source H.264 codec

Cisco to release an open-source H.264 codec

Cisco to release an open-source H.264 codec

Cisco to release an "open-source" binary-only H.264 codec

Cisco to release an "open-source" binary-only H.264 codec

Cisco to release an "open-source" binary-only H.264 codec

long periods of non-assertion can errode patent rights