Binaries and copyright and trademarks, oh my.

For the last few months there has been an ongoing murmur about Canonical’s intellectual property assertions. What I find interesting is that while people accuse Canonical of violating the Open Source Ethos they say nothing of other companies. Before I go any further, I need to state that I am not a lawyer and have not played one on TV. The thoughts and opinions are nothing more than my thoughts and opinions.

Red Hat
Here are some of the claims Red Hat makes in regards to intellectual property:

2. Intellectual Property Rights. The Programs and each of their components are owned by Red Hat and other licensors and are protected under copyright law and under other laws as applicable. Title to the Programs and any component, or to any copy, modification, or merged portion shall remain with Red Hat and other licensors, subject to the applicable license. This EULA does not permit you to distribute the Programs or their components using Red Hat’s trademarks, regardless of whether the copy has been modified. You may make a commercial redistribution of the Programs only if (a) permitted under a separate written agreement with Red Hat authorizing such commercial redistribution, or (b) you remove and replace all occurrences of Red Hat trademarks.


I think that clearly asserts that a copyright over programs. I will make the assumption they mean both the binary form and source code. They also prohibit redistribution unless all of the trademarks are removed unless you enter in to an agreement with Red Hat or you remove and replace all trademarks.

Here are some of the claims made by Canonical:

You can redistribute Ubuntu in its unmodified form, complete with the installer images and packages provided by Canonical (this includes the publication or launch of virtual machine images). Any redistribution of modified versions of Ubuntu must be approved, certified or provided by Canonical if you are going to associate it with the Trademarks. Otherwise you must remove and replace the Trademarks and will need to recompile the source code to create your own binaries. This does not affect your rights under any open source licence applicable to any of the components of Ubuntu. If you need us to approve, certify or provide modified versions for redistribution you will require a licence agreement from Canonical, for which you may be required to pay.

The disk, CD, installer and system images, together with Ubuntu packages and binary files, are in many cases copyright of Canonical

Canonical owns intellectual property rights in the trade dress and look and feel of Ubuntu (including the Unity interface), along with various themes and components that may include unregistered design rights, registered design rights and design patents, your use of Ubuntu is subject to these rights.

From what I read here anyone can redistribute Ubuntu in unmodified form, but if you are going to modify it and retain the trademarks you must seek approval from Canonical. You can redistribute it if you remove all trademarks and recompile. If you need Canonical to approve, certify or provide modified versions for redistribution you will need a license agreement.

The Case of Mint:
I want to remind everyone that this is my opinion only and that I am not a lawyer, nor have I played one in a movie. If I am to take the case of Linux Mint and apply what I have read from both Red Hat and Canonical what would Mint have to do? What I understand Mint does:

  • Uses Canonical’s repositories
  • Specifically claims that it is based on Ubuntu (and Debian)

If Mint was based on Red Hat® Enterprise Linux® they would need to remove all Red Hat® trademarks or enter in to a license agreement with Red Hat®.

In the actual case of Mint being based on Ubuntu they have the option of removing all Ubuntu trademarks and recompiling, or entering in to a license agreement with Canonical.

The Difference:
The only difference I see is that Canonical has explicitly claimed that one would have to recompile binaries. I am not a package guru nor a kernel developer… but I am under the impression that in order to remove trademarks and copyrighted art that the binary would need to be recompiled. This leaves me wondering what all the noise is about since it would appear recompiling would be necessary in the case of a distribution being based on Red Hat®.

One more thing:
To date I think Canonical has an outstanding record of working with redistributions of their software. They encourage such endeavors and want them to grow.

This entry was posted in FOSS, Linux, Ubuntu. Bookmark the permalink.

36 Responses to Binaries and copyright and trademarks, oh my.

  1. Pingback: Charles Profitt: Binaries and copyright and trademarks, oh my. | Hi-tech news

  2. This is the bit I don’t see: How does recompiling affect trademarks? Most appearances of Ubuntu trademarks would be in things like separate image and text files–the kind of stuff that appears on the menu or on wallpapers or in the Ubuntu software manager or what. And Mint is replacing all that stuff anyway. Most of the binaries in Ubuntu are just the usual libraries and programs, stuff that Ubuntu themselves got from Debian and in some cases tweaked a bit. So, what is the difference between Mint (for instance) using a binary of GlibC they got from Ubuntu, and Mint (or whoever) getting a source file of GlibC from Ubuntu and compiling it? How does the former somehow use an Ubuntu trademark but the latter not do so?

    I don’t mind Canonical asserting their right to their trademark; it’s just like CentOS can’t call their version Red Hat. That’s cool. It’s their tendency to be aggressively confused that’s annoying. Before they assert their rights it would be nice if they had a decent account of what those rights are and why compiling affects them.

    • Alan Bell says:

      “Most appearances of Ubuntu trademarks would be in things like separate image and text files–the kind of stuff that appears on the menu or on wallpapers or in the Ubuntu software manager or what. And Mint is replacing all that stuff anyway.”
      which is fine, so Mint should host their own repositories with all the replaced stuff and all the non-branded debs which they can optionally rebuild.
      “So, what is the difference between Mint (for instance) using a binary of GlibC they got from Ubuntu, and Mint (or whoever) getting a source file of GlibC from Ubuntu and compiling it? How does the former somehow use an Ubuntu trademark but the latter not do so?”
      no difference at all, and they can do that. They only have to rebuild the stuff with the trademarks, the rest they can grab source and binaries and rebuild or not rebuild, it makes no difference. The bit about rebuilding is in the same sentence as the stripping out trademarks bit. What they can’t do is point at the Ubuntu repos and distribute the binaries from there. Mint are in the wrong by making a commercial operating system they are pitching to OEMs with ubuntu repositories in the sources.list They just shouldn’t do it, they should sort out their own infrastructure if they want to do that.

  3. Mat says:

    Nice lucid description. One thing that the Canonical license doesn’t make explicit is do you need to recompile ALL the source code or just the source code for the packages that contain the trademarked material? In the case of Redhat, it seems to be all the source code, because CentOS I don’t think touches Redhat repositories. For Ubuntu, on the other hand, I think other distros still draw on the Ubuntu binaries for everything else. Do it the CentOS way, GPL is served, and there should be no issues all the way around — except for other distros needing to do the heavy lifting of recompiling everything.

    I think Debian has some other distros, that use their binaries, and Fedora too (after removing trademarks). But the ones I have used, Crunchbang, Korora, to me are really aimed at lowering the entry bar for the parent distro, and the parent distros are not commercial enterprises.

  4. Michael says:

    I think the biggest difference that RH never claimed that the RH name meant “humanity toward others” and this kind of affirmation that could confuse the community on who own the trademark, and I think that’s the root cause.

    Of course, Canonical is right to defend the trademark and to care of that as a steward of the community, but the messaging around it is likely confusing for some people . And that’s why Canonical get critized for stuff that others do without issues. You can compare to the ruckus of the Ubuntu One name. Canonical was perfectly able to do what they wanted from a legal point of view, but part of the community disagreed because they tought the Ubuntu name would be “non commercial and to the community”.

    • pleia2 says:

      RH never claimed that the RH name meant “humanity toward others”

      This is an excellent point that has really made me struggle with this entire discussion.

      I do agree that it’s important to protect trademarks and that’s part of Canonical’s role within the community, but I do place higher expectations upon them because of the philosophies that the Ubuntu project sought to embody.

  5. skugler says:

    How is recompiling binaries in general necessary to remove copyrights and trademarks?

    Binaries are clearly a derivative work of the source, the GPL and other copyleft licenses forbid additional restrictions. These trademark policies are thus in violation of the existing license.

    • Charles Profitt says:

      I am not sure… I do not package or compile. I do know that the programming I did in the past that art and other resources were embedded in the code. To remove the art (copyright) or trademark items the code I made would have had to be modified and recompiled.


      FSF and its founder, Richard Stallman, have expressed several views, some contradictory, on the derivative work test. Officially, the FSF acknowledges that the legal definition of a derivative work is not yet well defined as applied to software.

      I am not sure how all this plays out… as a non-lawyer there are a lot of moving parts. I don’t know how Red Hat prevents redistribution of RHEL, but it does it in a manner that would prohibit Mint for basing itself on RHEL.

      • “FSF and its founder, Richard Stallman, have expressed several views, some contradictory, on the derivative work test. Officially, the FSF acknowledges that the legal definition of a derivative work is not yet well defined as applied to software.”

        Do you have a citation for this?

      • Also I’ll add that Kubuntu’s Project Lead as already pointed out that distros like Mint do not need a license

        Also when I discussed this with Richard Stallman last year he said the FSF was investigating Canonical’s claim that people needed a license to distribute binaries containing GPL licensed works.

        • Charles Profitt says:


          You should note that Kubuntu’s project lead is neither a lawyer or an employee of Canonical authorized to speak on their behalf in regards to what a distro does or does not need. I assume, since Richard did not immediately have an answer to your question, that there is an actual legal question about it.

          • Kubuntu’s project lead as of last time I checked has had legal discussions with Canonical on the topic and its important to point out that the GPL was not written by a lawyer. I think Kubuntu’s Project Lead has significantly more expertise on the topic of licensing then you or I do and I respect that expertise.

  6. I don’t see how recompiling is necessary to remove trademarks. (Assuming it’s not built into the package, which is a special case.)

    Binary packages are clearly derivative works, so they fall under the GPL (or other copyleft licenses that apply). This means, putting additional restrictions on it is in violation with these licenses.

    Claiming to be a derivative is not a problem in trademark law. You must not claim that you “are something trademarked”, but using a trademark to describe the actually trademarked “thing” is entirely fine, not even the trademark holder can do anything against it.

    The bottom line to me is very clear: The policy put up by Ubuntu’s community council is legally highly questionable, morally wrong and in violation of copyleft licenses.

    • Alan Bell says:

      Rebuilding the package is necessary if it contains trademarks you want to remove. I don’t see how that is in any way a complicated concept. If they are in a binary then yes, a recompile would be neccessary, if they are not in a binary you could pick apart the .deb file and extract them manually, but it would be better to change the source package and rebuild it which will, as part of the process of creating the new .deb recompile any binaries.
      The bottom line to me is that Mint should host their own repositories and not point their users sources.list file at the Ubuntu repositories. Quite why Canonical even bothered to be nice enough to offer an agreement to allow them to carry on doing what they are doing is beyond me. It appears to have been some exceptional attempt to be excessively accommodating by Canonical, it would have been better for all if they had just said no, you can’t point Mint desktops at the Ubuntu repos.

      • PaddyD says:

        “The bottom line to me is that Mint should host their own repositories and not point their users sources.list file at the Ubuntu repositories.”

        Do Canonical host _all_ their own repositories? and not rely on public mirror’ing ?

        From a post by you further on up the page;

        “Mint are in the wrong by making a commercial operating system they are pitching to OEMs with ubuntu repositories in the sources.list They just shouldn’t do it, they should sort out their own infrastructure if they want to do that.”

        I rather suspect this is way way way closer to the real cause of the problem, I just hadn’t seen it posted so clearly before.

  7. David Stevenson says:

    I find comparing Canonical with Red Hat interesting, how would a compare with Debian work out?

    • Charles Profitt says:

      Since Debian is not a commercial company I did not think the comparison was valid. In the case of Canonical v. Red Hat things are different, but much closer. Canonical supports Ubuntu development and markets Ubuntu commercially. Red Hat support Fedora development; which, in turn supports Red Hat Enterprise Linux development. Red Hat markets RHEL commercially. I am not a intellectual property lawyer, but I assume there is are similarities. When you add in the existence of CentOS as a RHEL derivative (they recompile binaries and remove all trademarks) it becomes a very good comparison to Ubuntu and Mint. Recently, though CentOS became part of the Red Hat family. Which might make CentOS more like Kubuntu, Lubuntu or Xubuntu than Mint.

      • Jef Spaleta says:

        I’m still trying to wrap my head around one sticking point.
        Mint is point users to public repositories of ubuntu packages.. they are not actually distributing those ubuntu binaries as part of any install image or media.. right? Mint isn’t claiming their operating system on is Ubuntu right? They are just making it easy for people to install Ubuntu software that has Ubuntu trademarks inside of it. I don’t fully understand how pointing people to a repository of unmodified packages from which to choose to install addons beyond what is provided by the base OS that you installed…runs afoul of trademarks at all.

        I think there is precedent here in how say Ubuntu’s adobe flash package works to grab the flash binary from adobe’s distribution point to avoid actively redistributing the binary.

        So what’s the difference? Why can’t Mint make it easy for users to install unmodified Ubuntu branded software on to a Mint system if Ubuntu can make it easy for users to install unmodified Adobe software on to their Ubuntu system?

        • Charles Profitt says:


          I have difficulty understanding everything as well. I am not sure we even know if the issue is binaries in particular or not. I do know that when I did a test install of Linux Mint that there were several binaries that were installed — from the image that I downloaded. The install also had several files labeled as Ubuntu. As for your analogy of Adobe and Ubuntu… Adobe flash is a self-contained program and Ubuntu is simply providing an easy way to install Adobe’s binary. Ubuntu is not a binary program. Mint is not distributing it unmodified… they are modifying it.

          • Jef Spaleta says:

            I think you need to be very careful with the argument about Ubuntu being a single aggregate thing. I do not think that holds up at all considering the nature of how repositories work and interoperable considerations for binaries. The system is designed with binary modularity in mind. It suddenly no longer Ubuntu because I install gtk out of a random apt repo and replaces some of Canonical’s packages? It’s suddenly no longer Ubuntu because I’m running a vanilla kernel not vetted by Ubuntu QA? C’mon. That’s crazy. Historically Ximian did exactly that sort of thing.. layered over both red hat and suse back in the day. Gnome 2.x desktop packaging which replaced RHL packages but Ximian was a full distribution in itself. It was RHL with Ximian Desktop. Not a confusing situation at all for me as a user at the time and I don’t remember anyone crying trademark foul.

            Or are you saying that only Canonical trademark’d deliverables are able to pull from the aggregate repository? I think that’s a bridge to far, and I’ve never seen anyone else look at debian repository collections in that way. If that’s the interpretation Canonical wants to champion legally, its going to have a very chilling effect on Ubuntu derivative work more generally.

            And I very very much doubt that package names are even filenames are going to hold up as confusing under trademark scrutiny. For example /etc/redhat-release is a filename symlink that CentOS has always provided on system for backwards compat functionality as some software will read the contents of that file. But the filename itself not considered a wordmark for trademark law purposes. As filenames have functional interoperational value and you can’t just change the name of a file willy-nilly without breaking some existing software.

            And moreover fedora at the project level has tried to identify exactly what packages are in the repository which contain the Fedora trademarked material and provide explicit guidance to unlicensed derivatives that those particular packages should be replaced. It’s not just a mandate from on high that you have to treat the entire repository of binaries as Fedora. No, its a very specific mandate to protect the trademarks from dillution. And Fedora tries to make it possible for derivatives to work the the project to protect the value in the marks without unduly restricting their ability to use the component software binaries under the copyright terms for those individual binaries.

            Fedora even goes so far as to provide a generic-logos package which can be used as a drop in replacement for fedora-logos to try to prevent unmet packaging system dependencies for unlicensed derivatives who are working in good-faith to meet the trademark policy by excluding the fedora-logos package in their derivative. Fedora knows which derivatives are working in good faith, because they’ve installed generic-logos, or some other package, which conflicts with fedora-logos, as part of their base image. Not saying that covers all the trademark policy cases, but it covers a lot, and Fedora is trying hard to to have the trademarks be a block on using other binaries under the copyright licensing terms appopriate for those binaries. Portraying the whole repository as an aggregate of binary packages under the same licensing terms would be at best intellectually dishonest and at worst a violation of the licensing terms for some components.

            If Ubuntu has a serious trademark issue here with a derivative, meeting them halfway and choosing to implement a similar packaging solution to make it easier to strip the trademarks out, would seem like the… human and friendly thing to do. Letting the lawyers off the chain to attack was is arguable a very successful derivative…not as friendly for a project that regard the CoC as one of its greater social achievements. But in time you’ll figure all this out. If Canonical presses its trademarks against Mint at all any moving forward, with making any packaging adjustments to make it possible for Mint to avoid giving those marks to users, its going to cause more blow back and burn up even more good will.

          • Charles Profitt says:


            I made not claim that Ubuntu is a single aggregate thing. I, in fact, said it was different than Adobe flash. To be honest though I am not sure if Adobe Flash is a self-contained program or a collection of programs. I guess the point I was getting at was that Ubuntu allowing Adobe’s flash player to be installed as Adobe packaged it and intended for it to be installed is different than picking pieces from a distribution and then building a new distribution.

            To be honest I am not sure what the specifics are of the Mint issue. I also assume the generic statement on the Ubuntu site does not claim ownership on all the files contained in the distribution… it claims ‘some’ of them… just as the statement actually says.

          • Mat says:

            It’s been a while since I’ve had mint installed, but I suspect that their distribution requires several (likely lots of) packages in ubuntu repositories in order to meet its minimum released functionality. The easiest way to tell is to do an apt-get update after installing and see if it hits ubuntu repos. Ubuntu is self-contained in using its own repositories, anything else that is installed from other repos is additional functioanality for individual users (just verified on a new xubuntu 13.10 installation)..

          • Not sure how your points change anything.
            Just wondering–would the Ubuntu people actually be happier if Mint were taking Ubuntu packages, changing the filename to say “mint” rather than “ubuntu” and distributing that? I suspect they’d then get annoyed that the Mint people weren’t giving credit where it’s due–even though I’m sure many of the Ubuntu packages are in fact Debian packages with the “debian” in the filename changed to “ubuntu”.

            For that matter, I wonder how using actual Ubuntu packages and calling them Ubuntu packages can violate Ubuntu’s trademark. It’s like if I buy a bunch of cans of Coke and sell them for a markup; nobody expects me to erase the “Coke” brand name from the cans–they are actual cans of Coke. Ubuntu packages, unlike cans of Coke, just happen to cost zero dollars to acquire.
            Now if the Mint people made up their own packages and called those “Ubuntu”, that would be a violation of trademark.

          • Jef Spaleta says:

            In reply to Purple,

            package names and not necessarily protected items. And I do not think there is a precedent yet to show that they are. For example CentOS replaces the trademark protected content of the package redhat-logos… but still ships the package as named. They do not take the additional step of providing a replacement package called centos-logos and integrating into the packaging dependency system. The package name stays the same, but the contents differ. And if Red Hat’s legal eagles are okay with that, then its arguable that Canonical’s legal team should also be okay with that. But there is no legal precedent as far as I am aware…so its arguable. But I daresay it’s going to be hard for Canonical to show that package name causes confusion in the market place in a trademark sense. I look forward to them bringing such a claim to court.

  8. Mat says:

    Here is another comment on binary builds, actually a question, since I don’t know the answer: is everything in the process of taking source packages and hosting binaries on a server a GPL process? That is, are companies that create and host binary distributions/packages, obligated to share all the details of their build, QA, etc process? It seems to me there is a lot of intellectual property, time, and horsepower also embedded in how the build process is set up to verify quality packages.

    Going with the Redhat example, I believe it took CenOS quite a bit of effort to take Redhat source packages and produce their own binary repositories. In fact, with Redhat 6, CentOS took a lot of heat for it taking so long and the process, including QA, being so opaque to the community.

    • Charles Profitt says:

      I am not sure either. I know CentOS also took about two months to release RHEL updates… I assume they were required to do so by some mechanism, but I am not sure what Red Hat did to force that.

      • Jef Spaleta says:

        Let’s be very very clear about CentOS…. They rebuild from source because Red Hat does not make the binaries publicly available. Red Hat does not use a public mirror system to distribute binaries and this one fact greatly constraints how one can choose to build a derivative.

        What CentOS is compelled to do is replace the contents of redhat-logos package. They aren’t even compelled to rename the package! They still ship packages with redhat in the name. They still ship files on the filesystem with redhat in the name. They still ship binaries tools with redhat in the name (redhat_lsb_init for example). These are functional interoperational elements of the system. Filenames on the system are not covered by trademark. What is covered by trademark is in the redhat-logos package and CentOS replaces that content with appropriate CentOS branding. Nothing in the Red Hat trademark policy compells them to recompile the entire distribution of packages.

        Moreover, Fedora’s packaging is constructed in a similar way, putting the protected trademarked material that identifies Fedora project deliverables in the fedora-logos package. But Fedora does have public mirrors! Because of those public mirrors, Fedora fully expects people to created unlicensed derivatives that mix and match Fedora binary packages with externally produced binaries! In fact there’s no way to stop that from happening for many of the components in the Fedora repository. GPL licensed code very specifically gives users the right to reuse binaries from one aggregate collection in another aggregate collection without requiring the GPL covered binaries to be recompile as part of the re-aggregation. But Fedora does have a compelling need to protect its trademarks. Fedora is able to accomplish trademark protection without requiring all derivatives who want to reuse Fedora built binaries to submit to a license and avoids running afoul with GPL licensing requirements. Fedora does not impose any additional requirements on GPL licensed components in the Fedora repository aggregate collection of binary packages.

        What Fedora mandates is that unlicensed derivatives not use the Fedora trademarks..which means not shipping or using the contents of the fedora-logos package in the derivative work. That’s it. Nothing more. No hand waving arguments about the value or quality of the aggregate. Just don’t use the trademarked material in the fedora-logos package and you can reuse all the other pre-compiled binaries built by Fedora that you want (in the scope of their copyright licensing of course).

        Fedora even provides an alternative generic-logos package specifically to make it easier for binary consuming derivatives to avoid having to recompile ANYTHING. You can right now, make a derivative live cd which mixes Fedora repostiory binary content and external binary content..cleanly…using kickstart templates.. by simply excluding fedora-logos package and using the generic-logos package as provided in the Fedora repository instead. Boom binary derivative that does not dilute the Fedora marks internally. No requirement at all to recompile Fedora built packages. You just have to not include fedora-logos package if you choose to to recompile.

        Any policy for any distributor, such as Ubuntu, which tells you you cannot make a derivative work which re-uses publicly mirrored binary builds of the kernel, built by the distributor in question, is a straight up falsehood by the distributor. Similarly for gcc and for gtk and for many other components that are istributed under the terms of GPL. Ubuntu/Canonical cannot compell derivative authors into signing any trademark or ip agreement simply for reusing Ubuntu built binaries of many of the components in the aggregate collection of the Ubuntu repositories. At best Canonical has some authority to control the licensing under which Ubuntu distributes Canonical controlled projects, that Canonical has licensing authority for. But Ubuntu would have to explicitly note that they are no longer distributing those items under the terms of the GPL. But regardless, no repository wide IP policy can restrict reuse of most of the GPL code that Ubuntu is shipping, without being in breach of the GPL itself. I fully expect kernel authors and FSF to raise objections to Canonical’s IP policy as it applies to the full aggregate collection.

        • Charles Profitt says:


          It seems you have made a great number of assumptions in regards to what Canonical has or has not asked for. To date I am unaware of Canonical asking for a license on the kernel. In fact, I do not know what they have asked for in the license. Neither Canonical nor Mint have made those items public or disclosed them to me.

          Given Clems actual quote on the subject it would appear that a great many people, you included, have jumped to a set of worse case conclusions. I fully expect reality to be significantly different than what you have laid out.

        • Charles Profitt says:

          Despite the Red Hat binaries not being publicly available… from what you are telling me they could not stop someone for redistributing them based on the GPL. So it would take a grand total of one person getting a copy and then redistributing them… is that accurate?

          • Jef Spaleta says:

            That one customer, who chooses to make red hat built packages public to me or you, would no longer receive updates from Red Hat and would most likely void all their existing services contract with Red Hat including engineering support or things like that. It might not be worth it to the customer to lose the services they are contractually paying for. But if they did decide to make binary packages public, they could, as long as the packages were devoid of red hat trademarks. So no branded install media for example.

            In fact this was exactly what happened under the RHL model before RHEL. There were public mirrors of the binaries. Nothing has really changed, other than Red Hat tying continued access to services to the condition that customers agree not redistribute the binaries. If customers redistribute binaries under the RHEL, Red Hat terminates their access to services (including updates)…and the the binaries are still out in the public and could be remixed into a derivative.


          • Charles Profitt says:

            Thanks for that explanation… so RHEL depends on an implied indirect restriction on redistribution. In the end, regardless of method, they are restricting the redistribution of the binaries.

          • daveb says:

            Not on trademark / copyright grounds, no. The contract explicitly mentions that you have the right to redistribute binary packages via the open source license – doing so however violates your subscription contract. See RHEL license agreements.

          • Jef Spaleta says:

            I strongly disagree with your interpretation of the RHEL situation.

            Regardless, it doesn’t matter. As soon as GPL licensed binaries built by Red Hat become publicly mirrored anywhere, the can be made part of derived works as long as those derived works do not contain Red Hat trademarks. Red Hat will have no grounds under either trademark nor copyright law on which to prevent anyone from using such publicly available binaries in a new aggregate product if such binaries do not contain trademark protected materials. The fact that Red Hat builds and QA’s such binaries does not imply any special rights to Red Hat to prevent such binaries from being re-used or re-distributed as part of derived works as long as those derived works are marketed and branded distinctly as new aggregate products in the market.

            Ubuntu binaries are publicly mirrored, and thus the subset of binaries in the publicly mirrored Ubuntu archive devoid of Canonical trademarked material and distributed under the terms of the GPL (and other OSI approved licenses actually), can be re-used in a new derived aggregate work, without requiring recompilation from source. Such derived aggregate works, which contain some binaries compiled originally by Ubuntu and meant for Ubuntu licensed products, cannot be compelled to enter into an additional licensing arrangement with Canonical under trademark law. All that is required is that these new derived works market and brand themselves distinctly in the market place as a new aggregate work. The fact that some or even most of the components were original built for use in Ubuntu licensed aggregate products is immaterial. Nor does trademark protection require Canonical to seek such licensing. Canonical can seek such licensing for business reasons, but its not required under trademark law as part of due diligence to protect the marks.

            I really look forward to Canonical butting heads with the FSF over this in the upcoming six months, to clear this up a bit more. But in the meantime, the Ubuntu project could do a lot in terms of packaging policy to make trademark enforcement cleaner. I’ve sketched out how Fedora tries to minimize the problem space with packaging policy. That’s the best I can do in terms of constructive proactive mitigation to the problem.

          • Charles Profitt says:


            You can strongly disagree, but that makes it just a matter of semantics.

            A customer who purchases a Red Hat subscription is prevented from redistributing them because of the threat of losing their subscription.

            I certainly respect the efforts that Fedora takes to make trademark enforcement cleaner. I appreciate the discussion as it helps us all understand things better.

          • These are matters of law. The whole ball of wax is “just” a question of semantics.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s