GPLv3 pushed no one away [LWN.net]

GPLv3 pushed no one away

Posted Jul 25, 2013 8:33 UTC (Thu) by coriordan (guest, #7544)
In reply to: Some of us knew this 30 years ago by khim
Parent article: Android 4.3

> GPLv3 plugs this loophole. Unfortunately result was opposite to
> FSF's expectations: instead of opening the devices GPLv3 killed GPL.

Actually, the companies complaining about GPLv3 had already parted
company with free software by blocking the running of modified
versions. The freedoms of the GPL had already been killed. The
decision FSF was left with was whether these non-free companies should
be able to continue to use GPL software in their non-free DRM devices.

In this context, FSF's decision was obvious.

Those non-free companies try to place the blame on FSF (with some
success, unfortunately), but they actually abandonned free software
before GPLv3 existed.

to post comments

You can say it all you want, it doesn't make it true

Posted Jul 25, 2013 10:12 UTC (Thu) by oshepherd (guest, #90163) [Link] (10 responses)

Companies like TiVo aren't the ones who complained about the GPL3. How many notable open source contributions have TiVo made? Certainly none of them have ever made it to my news feed... Whatever complaints they may have had, they fell upon deaf ears

No, it was companies like Apple. Apple, who had a copyright assignment agreement on file with the FSF for their branch of GCC. Does that sound like a company that hated the GPLv2 to you? It doesn't to me. To me, it sounds like a company who were OK with the GPLv2; who, certainly for "infrastructure" pieces of software were absolutely fine working under that license and, as mentioned, assigning the copyright on their GCC works to the FSF.

The GPLv2 was a license that everyone, even the BSDs could begrudgingly agree on. The GPLv3, in a lot of people's world views, was a step too far; and I think you'll find, for a lot of people and companies, it's not the anti-"tivotization" clause that is the problem: it's the anti-patent-licensing clause.

I've said it before, and I'll say it again: the GPLv3 was the FSF's gift to the BSDs. Today, they're approaching having a GPL-free base system. 10 years ago I don't think anyone expected that this would be the point we were at today.

You can say it all you want, it doesn't make it true

Posted Jul 25, 2013 12:17 UTC (Thu) by Jan_Zerebecki (guest, #70319) [Link] (7 responses)

Apple hated the GPL since the beginning. They were the first to violate the GPL. They changed GCC to implement Objective C but couldn't get away with not distributing the source. Starting on slide 7 of that presentation http://ebb.org/bkuhn/talks/LinuxTag-2011/compliance.html more details are explained.

I don't think the fact that they had a copyright assignment for GCC in place contradicts their hatred for the GPL. IMHO it just shows that the GPL was successful for some time in forcing them to contribute once they got caught violating it. That is until they had the resources to rewrite the software they needed from scratch.

You can say it all you want, it doesn't make it true

Posted Jul 25, 2013 18:20 UTC (Thu) by oshepherd (guest, #90163) [Link] (6 responses)

If Apple were so hell bent against the GPL2, why would they go through the effort of signing a copyright assignment agreement with the FSF?

You portray things as if Apple/NeXT were hell bent against the GPL and contributing to GCC - if they were that hell bent, why did it take them 18 years (From the Objective C compiler's release in 1989 to Clang being open sourced - by Apple) to even start replacing it? Why did it take them until the dawn of the GPLv3 to begin heavily contributing to LLVM? Why do they, to this day, ship (GPLv2) versions of GNU Coreutils, Bash, Make, etc, if they despise the GPL so, when the BSDs could give them viable equivalents?

Lots of GPL zealots try to make it out that Apple are some sort of open source pariah; yet they have invested lots of effort into LLVM and Clang with no requirement that they release their work as open source. They have invested lots of work into WebKit, with no requirements that they support any platform besides their own, and yet it has become the worlds' most ported browser engine under their watch. They continue to give away things like libdispatch (Apache2), their variant of the DCERPC library (BSD), the ALAC codec (Apache2) and their CalDav/CardDav server (Apache2) without anybody so much as asking

Yeah, I have issues with Apple too. I think it goes without saying that they're definitely no saint, but it's disingenuous to claim that they hate the GPL.

I can think of far, far more toxic companies. Oracle spring to mind.

You can say it all you want, it doesn't make it true

Posted Jul 25, 2013 20:53 UTC (Thu) by khim (subscriber, #9252) [Link]

They have invested lots of work into WebKit, with no requirements that they support any platform besides their own, and yet it has become the worlds' most ported browser engine under their watch.

That one is direct result if GPL. LGPLv2 to be precise. WebKit was born when people complained about lack of Safari's source code. And it was always separable from Safari exactly because Apple knew that they will be forced to release source code.

IOW: Apple (and many other companies) accepted GPLv2. They were not happy, but they accepted GPLv2 "tit for tat" principle. When GPLv3 arrived with it's "I'm altering the deal. Pray I don't alter it any further." approach they rebelled.

You can say it all you want, it doesn't make it true

Posted Jul 25, 2013 20:53 UTC (Thu) by Jan_Zerebecki (guest, #70319) [Link] (4 responses)

Signing the copyright assignment seems like a logical next step because it saves them effort in maintaining the fork. Liking or hating the GPLv2 doesn't enter into that equation because the code in question was already GPLv2 licensed.

AFAIK, GPLv2 Bash, Make and so on are not shipped in Apples mobile devices. I suppose those are just not that important nor substantially modified.

I don't think Apple wants to avoid the use of copyleft software at all cost. I do think that the Apple of today hates the general idea of being forced to publish their written/changed software by copyleft more than they did back then. I also do think Apple hates the GPLv3 more than the GPLv2, because the Anti-Tivo clause makes the GPLv3 incompatible with them locking down their mobile hardware.

> The GPLv3, in a lot of people's world views, was a step too far; and I think you'll find, for a lot of people and companies, it's not the anti-"tivotization" clause that is the problem: it's the anti-patent-licensing clause.

If the explicit patent-licensing clause were in general a problem for Apple why do they use the Apache2 license, which has a similar clause?

If Apple likes the GPLv2 why do they prevent developers from using it for mobile Apps in their Appstore legalese?

You can say it all you want, it doesn't make it true

Posted Jul 25, 2013 21:54 UTC (Thu) by Cyberax (✭ supporter ✭, #52523) [Link]

They are afraid to trigger the "derived work" clause and be forced to disclose at least parts of iOS source code.

You can say it all you want, it doesn't make it true

Posted Jul 26, 2013 14:52 UTC (Fri) by raven667 (subscriber, #5198) [Link] (2 responses)

> the Anti-Tivo clause makes the GPLv3 incompatible with them locking down their mobile hardware.

I think this was handled poorly by the GPLv3 design committee, a public relations failure as much as a technical legal one. ISTM the way to handle anti-TiVoization is to make the lock-down keys user-modifiable or provide a tamper switch to stop the boot check, the way that the Chromebooks work. You can either run the vendor-supplied firmware with DRM and whatnot or you can, with some small amount of effort, replace it with your own without the DRM.

The whole thing where it was said that you needed to divulge your private keys to make signed firmware images was a huge black eye and has done much to prevent uptake of the GPLv3 by device makers. This perception should have been fought tooth and nail both with extensive propaganda and supporting changes to the legal text.

That's just my opinion though.

You can say it all you want, it doesn't make it true

Posted Jul 30, 2013 17:44 UTC (Tue) by rfontana (subscriber, #52677) [Link] (1 responses)

> I think this was handled poorly by the GPLv3 design committee,

The what?

> ISTM the way to handle anti-TiVoization is to make the lock-down
> keys user-modifiable or provide a tamper switch to stop the boot
> check, the way that the Chromebooks work. You can either run the
> vendor-supplied firmware with DRM and whatnot or you can, with some
> small amount of effort, replace it with your own without the DRM.

This is one way to comply with the so-called anti-TiVoization
provisions of GPLv3.

You can say it all you want, it doesn't make it true

Posted Jul 30, 2013 19:16 UTC (Tue) by raven667 (subscriber, #5198) [Link]

>> make the lock-down keys user-modifiable or provide a tamper switch to stop the boot check, the way that the Chromebooks work.

> This is one way to comply with the so-called anti-TiVoization
provisions of GPLv3.

Yes and this was poorly advertised and promoted. Even today people commonly claim that the only way to comply with the GPLv3 is to provide the private signing keys or that the GPLv3 is fundamentally incompatible with any kind of boot time checks. I believe that this is one of the main reasons why Linus and many Linux developers don't like the GPLv3 and why many large projects are GPLv2-only or don't pick a GPL-family license at all. I think that we now have less Free Software using the GPL-family after the GPLv3 than before its introduction, which seems like a disaster to me.

I don't see much other than the Google devices which even try to be open, and the Android team didn't even chose the GPL because they believe it is incompatible with their business interests. Whether that is true or not, the fact that the meme is out there and so strong it its own problem that should have been avoided.

You can say it all you want, it doesn't make it true

Posted Jul 25, 2013 16:34 UTC (Thu) by Frej (guest, #4165) [Link]

Considering that for Xcode they had to parse xml output from gcc to avoid linking... that would certainly be a motivating factor :). Also, GCC missed the idea of compiler integration/service for other tools instead of a simple batch processor. And for this GPL vs. LPGL matters quite a lot.

GCC wouldn't even consider any kind of plugins/library approach until LLVM came along.

GPLv3 and fragmentation

Posted Jul 31, 2013 8:03 UTC (Wed) by marcH (subscriber, #57642) [Link]

> The GPLv2 was a license that everyone, even the BSDs could begrudgingly agree on. The GPLv3, in a lot of people's world views, was a step too far; and I think you'll find, for a lot of people and companies, it's not the anti-"tivotization" clause that is the problem: it's the anti-patent-licensing clause.

> I've said it before, and I'll say it again: the GPLv3 was the FSF's gift to the BSDs. Today, they're approaching having a GPL-free base system. 10 years ago I don't think anyone expected that this would be the point we were at today.

Interesting theory, I'm looking forward to the next decade to see how this all ends up.

One thing you can't argue with is: whereas the GPLv2 is only a software license, the GPLv3 is a complete, "forced freedom" package. What about people and companies who only want a software licence?

Fortunately the FSF cannot remove the GPLv2, meaning it cannot remove people's and companies' freedom to use only the software licence. Unfortunately the fragmentation between GPLv2 and GPLv3 looks deadly for both.

I personally do not mind about tivoization. It still leaves me the freedom to buy from a competitor who copies all the GPLv2 code from TiVo. If there is no market for such a competitor because consumers all prefer tivoization (at least for the moment) then so be it. If consumers don't want software freedom then it should not be shoved down their throat.

I do mind silly software patents but again I don't like the idea of shoehorning a patent fix into a licence, these are different issues that should be treated independently.