LWN: Comments on "ZeroMQ and Crossroads I/O: Forking over trademarks"

ZeroMQ and Crossroads I/O: Forking over trademarks

pieterh — Wed, 06 Nov 2013 15:39:59 +0000

It's ironic that at the bottom of this page it says, "Linux is a registered trademark of Linus Torvalds".

Trademarks are important here yet not in the way presented. There were no requests from anyone to use the ZeroMQ trademarks in other products. It is strange to see that presented as justification.

The fork was simply over flatness of our playing field. Martin and Martin ("M&M") wanted special rights to release code that was buggy, incompatible, and broke user space, and call it "ZeroMQ". At the same time they felt justified in deciding the road map, rejecting patches, and imposing their vision over the codebase no matter what the cost to users.

There was an old agreement we made when we launched the project. iMatix funded it, and owned the name and domain, was responsible for the community, and by 2011, for official releases. That was my job: evangelism, documentation, community building, and making stable releases (no easy job when the master code kept changing dramatically). I've been the benevolent dictator of the community since the start, though didn't interfere in the process until there were problems to fix.

In 2011 we found ourselves with no less than four incompatible versions: 2.x, 3.0, 4.0, and then a 3.1 cowboy release, which M&M made while I was flying home from Seoul, after we'd spent a full week together there. At that point I reached out to Mikko, Ian, Chuck, and with their help codified the rather harsher process we use today. I told M&M, these are the rules now, accept them or leave - it is LGPL, after all.

A level playing field means all contributors have the same rights. The trademark is a coercion that enforces this rule. And that's it. Anyone can fork the code and apply a different process, but they cannot call that result "ZeroMQ".

How well does the process work? Clearly, it's not the only way to make software. If I was a large firm pumping money into messaging, to build a nice conventional support business, I'd use a BSD license and a playing field that tilted my way.

However to build a truly open community, and to make software that is accurate and stable (and look at the latest V4 release, how few issues it has, and yet how far it's come), it works.

Personally, I find the ZeroMQ community, since M&M's departure, to be lively, happy, and really pleasant to work with. We've lost all the stress of broken code and weird philosophical design debates, and instead we see smooth organic growth of the code. Master is almost perfect, almost all the time. Working on ZeroMQ is enjoyable, and I think that's a feeling all contributors share. Don't take my word for it, go ask on the zeromq-dev list or #zeromq IRC channel.

If you are making free software, take a look at our process: http://rfc.zeromq.org/spec:22. It's surprisingly effective, and for that I have to thank M&M sincerely, since they were the main motivation to write it.

The meanings of a fork

ncm — Mon, 30 Apr 2012 21:05:32 +0000

Each fork is an adaptation to a failure. Opinions will differ on the nature of the failure. The ability to fork is good. The fork itself may be good, under the circumstances. The circumstances are always unfortunate.

That's not to say that every project must avoid forks at all costs. Nothing made by humans can satisfy everybody. Failure to satisfy everybody is unfortunate but expected. If enough of those dissatisfied can work together on a fork to meet their own needs, that's a much better outcome than remaining dissatisfied.

ZeroMQ and Crossroads I/O: Forking over trademarks

Baylink — Sun, 08 Apr 2012 17:05:31 +0000

Sure, he could call it a 0mq Toaster Control Unit.

ZeroMQ and Crossroads I/O: Forking over trademarks

kragil — Thu, 05 Apr 2012 08:06:18 +0000

Forks are good.
It's evolution, baby.

ZeroMQ and Crossroads I/O: Forking over trademarks

chloe_zen — Wed, 04 Apr 2012 04:00:42 +0000

... because what zeromq needed most was more confusion. :-(

ZeroMQ and Crossroads I/O: Forking over trademarks

giraffedata — Sat, 31 Mar 2012 16:36:58 +0000

You seem to be talking about the name for the protocol, while the trademark in question is the name of a piece of software that implements the protocol. I don't believe it would satisfy Sustrik if the protocol were given a new non-proprietary name while iMatix retained the trademark on the piece of software Sustrik works to distribute.

The freedom to use the mark that Sustrik advocates wouldn't be the freedom for a developer to call his code "Foobar Enterprise 0MQ" to indicate that it uses the same protocol as the original 0MQ; it would be the freedom to call it that to indicate that it contains all the features of the original software.

ZeroMQ and Crossroads I/O: Forking over trademarks

xtifr — Thu, 29 Mar 2012 23:55:34 +0000

This isn't the first case where a project has been split purely over trademarks with no difference in the copyright license, and with the ability for patches to flow freely between the projects. Debian Iceweasel is the same. Of course, Iceweasel simply tracks Firefox, and imports any changes; there's no actual independent development going on on the Iceweasel side. Still, I think it's an interesting precedent.

ZeroMQ and Crossroads I/O: Forking over trademarks

iabervon — Thu, 29 Mar 2012 17:33:32 +0000

In general, things that become IETF standards have different standard names from the names they originally had, when a company has a trademark on the original name. For that matter, a standard has to permit competing implementations, which are obviously going to be not from the ZeroMQ community but confusingly similar to ZeroMQ, and therefore violate any sensible trademark policy on that name. Of course, "Crossroads I/O", while a fine project name, isn't any better for this purpose. If they want to have a standard, they'll need a generic name, like "facial tissue" or "copy machine" or "XMPP", and that's what other companies would claim support for without having conflicts with trademark policies on it.

ZeroMQ and Crossroads I/O: Forking over trademarks

wookey — Thu, 29 Mar 2012 11:38:17 +0000

<blockquote>You cannot create "Foobar Enterprise 0MQ" in the way that Red Hat have created "RH Enterprise *Linux*".</blockquote>

I don't follow the logic of this. 'Linux' is trademarked too, just like '0MQ'. Wide usage is allowed due to a liberal policy and light application of the power trademark law provides.

0MQ could be the same if it wanted. It's not entirely due to trademark law per se. Now perhaps the point is that iMatix/PeterH aren't prepared to chnage the policy at all, but they seem pretty reasonable to me in the linked communications.