LWN: Comments on "VMware Suit Concludes in Germany"

Podcast discussion from Bradley and Karen of Conservancy regarding this topic

bkuhn — Tue, 23 Apr 2019 00:51:07 +0000

For posterity, I'm adding a link here to Free as in Freedom, Epsiode 0x66, wherein Karen Sandler and I discuss details of the conclusion of the Christoph Hellwig vs. VMware case. You can download the faifcast in mp3 or ogg format.

VMware Suit Concludes in Germany

bkuhn — Mon, 08 Apr 2019 20:56:30 +0000

A few comments on various parts of this thread:

A lot more evidence than mere git blame was provided to the Court. Conservancy assisted Hellwig with this detailed analysis (which I did most of) and Hellwig included his own detailed analysis, too. Our understanding is the Court simply chose not to examine this evidence in detail, nor did they appoint an expert (not like experts in the USA system, more like a Special Master in the USA — an expert who works for the Court to help them understand the sisues), even though both sides wanted one appointed. This was an odd decision and was surely a factor in why the Court erred in their understanding of Hellwig's copyrights.
While this is obviously not an ideal ruling, the action has compeled VMware to comply — the only “loss” is they are not doing so in a reasonable timeframe, and it would have admittedly been nicer if the Court compeled celerity in those compliance efforts.
Either way, no precedent is set, as others mentioned. Germany isn't a common law legal system. The ultimate goal here was never to set precedent, but to compel VMware to DTRT and follow the requirements and terms of the GPL.
This case was much different than typical GPL violations on Linux, because VMware took only part of Linux to combine with their vmkernel; so fewer Linux copyrights were infringed than is typical in violations (and thus a fewer number of copyright holders were even able to bring an action). Much of the code in this subset that VMware misappropriated was not held by individuals, but companies. (Hellwig was smart to be sure to keep his own copyrights.) Red Hat (whom folks mentioned extensively in this thread) was indeed a large copyright holder in the copyrighted material VMware misapproriated, and Red Hat could have been a big help in this case but declined to assist (as did other copyright holders we asked).
Ultimately litigation is expensive and unpredictable, but is necessary in those last resort cases where companies are boldly refusing to follow the terms of the GPL. If no one ever sues over GPL violations, bad actors have no reason to comply, and that unfairly punishes the good actors who comply willingly and give up the opportunity for proprietary competitive advantages. Every historical attempt to protect communities from companies via litigation has faced losses, wins, and everything in between before they ultimately succeeeded. Software freedom via copyleft is a long strategy that requires careful, long-term work; no one expects one single lawsuit will decide everything about copyleft. Any single violation matter is just a step toward software freedom for users. In my view, only a community-oriented organization that puts user freedom first can take these steps; companies, even if they're generally friendly to FOSS will always prioritize their business interests over social good.

VMware Suit Concludes in Germany

aggelos — Thu, 04 Apr 2019 13:19:19 +0000

Heise also mentions that when discussing the possibility of an out of court settlement; it's not entirely clear to me that the judge's statement was made in that same context, though it would make sense.

That said, judges can be very conscious of their role as defenders of the status quo (as they perceive it). In fact, judicial decisions that undermine perceived norms are (a) usually in response to shifting norms in the society at large / political discourse (b) a bit like comparable european paintings: notable enough to be talked about decades later.

VMware Suit Concludes in Germany

pebolle — Thu, 04 Apr 2019 12:06:30 +0000

> It was reported in a Heise article at that time, and it's literally what I wrote above.

That article says that the judge remarked that Hellwigs complaint apparently was made for "idealistic" reasons. So neither "unhappy" nor "ideological" seem accurate here.

And that the complaint was made for idealistic reasons doesn't strike me as an unreasonable thing to say for a judge in a case like this. What's wrong with a remark like that?

(Of course, I'm assuming Heise reported correctly here.)

VMware Suit Concludes in Germany

rgmoore — Thu, 04 Apr 2019 11:48:15 +0000

Redhat may not require code assignment, but they should be able to claim the contributions from all their employees. Since they're regularly contributing a substantial percentage of the code every time LWN gives the contribution statistics, they presumably can claim a fair fraction of the kernel under their corporate copyright.

VMware Suit Concludes in Germany

lkundrak — Thu, 04 Apr 2019 07:33:16 +0000

> Redhat doesn't require copyright assignment that I'm aware of

My employment contract with Red Hat has a copyright assignment clause.

It's entirely possible that some engineers have different contracts. Nevertheless, a quick "git grep 'Copyright.*Red Hat'" in kernel tree (assuming it's a reliable indicator) suggests that Red Hat still holds copyrights for significant parts of the kernel.

VMware Suit Concludes in Germany

cladisch — Thu, 04 Apr 2019 04:26:46 +0000

> the court decided that Cristoph's contribution to the stolen source was not significant enough to give him standing using VmWare's argument that it constituted less than 2% of the total stolen code.

This is not correct. The court did not say that Christoph's code is too small in relation to the entire code, but that Christoph's original contributions, when looked at in isolation (as if they were an independent program), are not significant enough to deserve copyright protection. And this was based not on the actual code, but on how Christoph described it in his pleadings.

VMware Suit Concludes in Germany

ILMostro — Thu, 04 Apr 2019 04:20:47 +0000

Ideological reasons for copyleft are what financial reasons for copyright are; i.e. understood value.

VMware Suit Concludes in Germany

cladisch — Thu, 04 Apr 2019 04:17:03 +0000

It was reported in a Heise article at that time, and it's literally what I wrote above.

VMware Suit Concludes in Germany

rahvin — Wed, 03 Apr 2019 21:13:16 +0000

Redhat doesn't require copyright assignment that I'm aware of, they'd need to get all their coders on board on the suit. This standing argument is just ridiculous, again, if you judge it the way the court did any large project of FOSS code could easily be stolen in Gemany unless you can get a significant percentage of the authors to back a lawsuit.

The court's logic was just plain stupid. Does this mean you can steal microsofts driver code as long as it's less than 2% of the total code base?

VMware Suit Concludes in Germany

aggelos — Wed, 03 Apr 2019 21:11:43 +0000

This is most interesting! Would you care to relay/paraphrase it here?

VMware Suit Concludes in Germany

Cyberax — Wed, 03 Apr 2019 21:08:35 +0000

A large company like RedHat/IBM can still sue.

VMware Suit Concludes in Germany

rahvin — Wed, 03 Apr 2019 21:07:20 +0000

Why? Apparently if the code stolen is large enough with a large enough group of authors no one has standing to sue in Gemany, so it's effectively already public domain.

VMware Suit Concludes in Germany

rahvin — Wed, 03 Apr 2019 21:04:43 +0000

My memory of the original verdict is that the court decided that Cristoph's contribution to the stolen source was not significant enough to give him standing using VmWare's argument that it constituted less than 2% of the total stolen code. And that the court used this standing argument to dismiss the case. The documentation issue was secondary in my memory of the lower court verdict, the decision turned on Vmware's argument about Christoph's code being insignificant in the whole. Maybe the appeal turning on other factors but the original court decision wasn't about how the information was presented IIRC.

The result being that if you still a big enough chunk of FOSS code you can apparently use it in Germany because no individual author will have standing to sue without gettting a large percentage of the code base's authors to step forward.

Vmware should have lost this case, they are using FOSS drivers in their commercial product.

VMware Suit Concludes in Germany

Cyberax — Wed, 03 Apr 2019 20:48:30 +0000

The kernel should just relicense itself under BSD to be fair to everybody.

VMware Suit Concludes in Germany

cladisch — Wed, 03 Apr 2019 20:45:04 +0000

> > And the court was not happy that the case was initiated for ideological reasons.
>
> Do tell. Are you refering to a part of the decision?

That was a remark during oral proceedings.

VMware Suit Concludes in Germany

aggelos — Wed, 03 Apr 2019 18:43:27 +0000

Another thing to take into account is that the lawyer representing Christoph in this case, was also the lawyer for a number of (mostly successful, AFAIK; at least one of them in the same regional court) GPL enforcement lawsuits filed in Germany on behalf of Harald Welte.

So it sounds like they have some experience regarding the acceptable standard of evidence in the german court system.

VMware Suit Concludes in Germany

aggelos — Wed, 03 Apr 2019 18:32:40 +0000

My german isn't great, but it seems the court was also given specific examples of code developed by Christoph, distributed as part of linux, that can also be found in ESXi.

Not that it matters that much; if what you say is true, given extensive documentation (and, AFAIU, its refusal to consult experts) the court could have easily replied that it's completely unreasonable to ask highly skilled legal professionals such as themselves to follow endless (perhaps machine-generated? :-)) instructions on how this part of this listing corresponds to that part of that listing and whatnot. There's usually no shortage of reasons the judiciary can put forward to refuse to examine the facts.

And the court was not happy that the case was initiated for ideological reasons.

Do tell. Are you refering to a part of the decision? If so, could you point everyone else to it?

VMware Suit Concludes in Germany

dunlapg — Wed, 03 Apr 2019 14:56:15 +0000

If that's accurate, I'm not that surprised that the court ruled the way they did.

Saying, "This code is mine; here's the git history, check it yourself" is a bit like submitting a patch to the kernel and saying, "This patch fixes a bug; the code is there, check it yourself." My experience with submitting patches to Linux is that they have a pretty high bar when it comes to commit messages -- demanding not only that the code be clear and correct, but that the commit message contain a very detailed description of the problem, the symptoms, and what the patch does to fix it. This matches my view is that a code reviewer's job should primarily be verification. The patch author should do the work of describing what the current situation is, why that's a problem, and what their patch does to fix it; the reviewer can then simply check to see that all of those things are accurate. It's not really the reviewer's job to reverse engineer all of those bits from the patch.

I'm willing to bet that the courts view themselves the same way: What they want to see is a detailed explanation of what code Christoph has written that ended up in VMWare's binary, and they'll verify that the explanation is valid. They don't think it's their job to generate the explanation themselves.

Given that, it seems like SFC missed an opportunity to establish precedent / best working practices in this area. Figuring out the best way to establish ownership would be a lot of work up front for the first person, but once there was at least one successful "template" to follow, the second person would have a much easier time of it.

VMware Suit Concludes in Germany

pizza — Wed, 03 Apr 2019 12:43:15 +0000

> 3) It has been proven that other companies can get away with it as well. No single developer, or even a small group of them, can enforce the GPL.

It's probably more acccurate to state that a single developer in a huge project can't enforce the GPL because they only "own" a tiny percentage of the overall codebase.

Smaller projects where individual developers hold the copyright to a much larger proportion of the code will fare better. (And for most projects, a handful of folks hold the copyright to everything..)

VMware Suit Concludes in Germany

cladisch — Wed, 03 Apr 2019 11:48:38 +0000

The court didn't specifically tell, it just said that what Christoph did is not enough:

He said "look at https//git.kernel.org/".
He provided an archove of the entire Linux source code.
He provided "git blame" outputs, and diffs of the original source files an VMWare's source files, but without connecting these two sets.
He did not argue (enough) how his modifications had a high enough level of creativity to be covered by copyright.

And the court was not happy that the case was initiated for ideological reasons.

VMware Suit Concludes in Germany

jezuch — Wed, 03 Apr 2019 10:49:55 +0000

The difference here is that in Common Law the courts have the power to create laws while Civil Law jurisdictions reserve this right to the legislative branch only; the courts may only interpret the law. Courts make decisions strictly on a case-by-case basis, but they definitely do look at decisions made by other courts in similar cases. Over time a consensus emerges and a "judicial interpretation" of the law is established (there are also other interpretations which have to be taken into account, one of them based on the presumed intent of the lawmakers, for example).

IANAL As Well of course

VMware Suit Concludes in Germany

mageta — Wed, 03 Apr 2019 09:52:22 +0000

> 3) It has been proven that other companies can get away with it as well. No single developer, or even a small group of them, can enforce the GPL.

And the bigger groups in our ecosystem have no interest in enforcing it.

VMware Suit Concludes in Germany

dunlapg — Wed, 03 Apr 2019 09:29:43 +0000

From Christoph's press release:

The requirements imposed by the court [to prove ownership] were extraordinarily high and make it very difficult for individual Free Software developers to assert their rights alone.

I'd be interested to see a description and analysis of what those requirements were.

VMware Suit Concludes in Germany

marcH — Wed, 03 Apr 2019 02:24:14 +0000

I bet precedence does matter in most civil law systems (especially from a high court) however it's nowhere near as sacred as in the US and much more rarely presented in courts. Judges will consider precedence but they will not automatically assume it was right and not necessarily feel bound by it.

VMware Suit Concludes in Germany

cornelio — Tue, 02 Apr 2019 23:20:23 +0000

Any way you look at it, the case was a loss for the GPL.

1) VMWare got away with it, they are changing the code only because they wanted to and they are not sharing anything.

2) SFC uselessly dumped lots of money and time into the german legal system.

3) It has been proven that other companies can get away with it as well. No single developer, or even a small group of them, can enforce the GPL.

Hellwig is not appealing because he has no case and would simply lose again.

VMware Suit Concludes in Germany

pbonzini — Tue, 02 Apr 2019 21:48:38 +0000

Well it's not just organizations but also audience, so it makes sense for them to point that out.

Also, the equivalent of the US Court of Appeals (e.g. in Italy the Corte di Cassazione) can set a "weak precedent" even in Civil Law systems, because their sentences are usually taken into account by lower courts even if strictly speaking they don't have to.

VMware Suit Concludes in Germany

marcH — Tue, 02 Apr 2019 16:51:28 +0000

> this ruling does not set precedent and the implications of the decision are limited.

IANAL and especially not a German one but it seems that organizations from English speaking countries regularly overestimate the importance of precedence in Civil Law systems.