LWN has an interesting article "Permissive licenses, community, and copyleft" https://lwn.net/Articles/660428/. It is about a talk by HP's Chief Technology Officer Martin Fink:
He then ended the session with an extended appeal to move the open-source software industry away from permissive licenses like Apache 2.0 and toward copyleft licenses like the GPL. Not doing so, he said, puts the FOSS community at just as much risk of collapse as license proliferation threatened to in years past.
Regards, Matthias
Good link Matthias!
Great presentation, even when taking the pragmatic view. Apart from the community-statement, I'd like to highlight his statement that copyleft stimulates collaboration, whereas non-copyleft projects require committees and other bureaucratic organization-parts to avoid partners splitting off and doing their own thing, not contributing back as a result.
Quick URL to the video: https://www.youtube.com/watch?v=zxIEDNyZOkA Conclusions start around 24mins in.
I'm preparing a presentation for the Dutch T-Dose conference which includes this topic as well. Glad other people are noticing the same symptoms. http://www.t-dose.org/2015/talks/open-source-has-won-now-what
Regards, Nico (NL)
On do, 2015-10-29 at 17:53 +0100, Matthias Kirschner wrote:
LWN has an interesting article "Permissive licenses, community, and copyleft" https://lwn.net/Articles/660428/. It is about a talk by HP's Chief Technology Officer Martin Fink:
He then ended the session with an extended appeal to move the open-source software industry away from permissive licenses like Apache 2.0 and toward copyleft licenses like the GPL. Not doing so, he said, puts the FOSS community at just as much risk of collapse as license proliferation threatened to in years past.
Regards, Matthias
Nico Rikken nico.rikken@fsfe.org wrote:
Great presentation, even when taking the pragmatic view. Apart from the community-statement, I'd like to highlight his statement that copyleft stimulates collaboration, whereas non-copyleft projects require committees and other bureaucratic organization-parts to avoid partners splitting off and doing their own thing, not contributing back as a result.
Quick URL to the video: https://www.youtube.com/watch?v=zxIEDNyZOkA
Thanks for the link. Pure comedy gold.
Given how much proprietary software HP is peddling, I think it's in a pretty poor position to lecture others about which free licenses they should use.
Quoting a LWN comment from fredrik: | So this anti-permissive rhetoric isn't in any way driven by any | ambition by HP to commoditize software, by blocking any opportunity | for proprietary extensions from competitors, on select markets to | shift customer demand towards their own hardware on those markets? | | HP, please let me know when you have licensed Integrated Lights Out | under the GPL. Oh, and the firmware for all those fine printers too. | I've always wanted to fix the bug that stops my printer from using | all of the ink in the cartridge. Go go GPL! | | Mr Flink also manage to conflates the Apache 2.0 license with the | organizations that prefer to use said license. https://lwn.net/Articles/660715/
If I understood Mr. Fink correctly, his argument is that HP does not open source products like HPUX (explicitly mentioned in the talk) because open source projects need a community to strive. Wait, what?
Mr. Fink also seems to knowingly conflate the Linux kernel with the free software community in general, for example when he spreads the FUD about DTrace and ZFS which according to him can't be used by "the community".
While the fact that they are licensed under the CDDL might be inconvenient for GNU/Linux distributions, for various other operating systems it's not a big deal and (from their point of view) even preferably to a more restrictive license like the GPL.
In somewhat related news, when I try to get additional information about OpenSwitch at http://www.openswitch.net/ I merely see an image of an arrow and no actual text on the website.
Looking at the page sources, the page is trying (and failing) to execute unsigned code on my system and the code does not seem to have a license header.
I hope the conference organizers can get Larry Ellison next year, I'm sure he has similarly valuable insights about free software licenses.
Fabian
Hi Fabian,
↪ 2015-11-01 Sun 12:20, Fabian Keil freebsd-listen@fabiankeil.de:
Mr. Fink also seems to knowingly conflate the Linux kernel with the free software community in general, for example when he spreads the FUD about DTrace and ZFS which according to him can't be used by "the community".
In that case, it's not far fetched considering the licenses for ZFS was apparently designed with the intention to make it incompatible with major existing licenses used in large shares by the free software community. (That's what I have heard repeatedly anyway.)
Free software licenses should reduce friction to allow greater flow of code, instead of deliberately increasing it.
That, and without forgetting that there are (were?) patents on ZFS stuff, which you cannot completely ignore (even outside the Linux kernel community).
While the fact that they are licensed under the CDDL might be inconvenient for GNU/Linux distributions, for various other operating systems it's not a big deal and (from their point of view) even preferably to a more restrictive license like the GPL.
The fact that there are more permissive licenses than the GPL doesn't mean it's not a big deal to make a GPL incompatible license.
It's a big deal to make a license that's not compatible with the most popular licenses, whether these licenses are copyleft or non-copyleft.
Best,
On Sunday 1. November 2015 21.41.47 Hugo Roy wrote:
Hi Fabian,
↪ 2015-11-01 Sun 12:20, Fabian Keil freebsd-listen@fabiankeil.de:
Mr. Fink also seems to knowingly conflate the Linux kernel with the free software community in general, for example when he spreads the FUD about DTrace and ZFS which according to him can't be used by "the community".
In that case, it's not far fetched considering the licenses for ZFS was apparently designed with the intention to make it incompatible with major existing licenses used in large shares by the free software community. (That's what I have heard repeatedly anyway.)
I would have to revisit the discussions and opinions about this from the fairly-well-known people who once worked for Sun. I don't think it was only my perception that various factions within Sun didn't want to see Linux enriched by code that supposedly gave Solaris an advantage. An impression did appear to build up over the years that Linux had eclipsed Solaris and that this was seen as undeserved.
Of course, the "success, but only on our terms" attitude that those factions nurtured ultimately resulted in the end of Sun and some of those people now having to work for Oracle. Similar inputs are presumably feeding the recently- reported malaise in the Apache OpenOffice project and some latent antagonism towards LibreOffice, where you get to see what happens when people are willing to drop the "only on our terms" bit and work with the community.
I personally think that Sun could have played a masterstroke by relicensing Solaris and accompanying technologies under GPLv3 or later. That would have shown up the GPLv2-only advocates in the Linux kernel community and possibly pushed Linux to become GPLv3-compatible. But as Hugo points out, there were patents on ZFS that the corporate hierarchy presumably wanted to withhold, and those factions presumably felt that all of this would have been "giving away the farm" or whatever. (One also wonders what influence some corporate contributors behind Linux have had in keeping it away from GPLv3 for similar reasons.)
Of course, relicensing a large, previously-proprietary product is a huge challenge. It arguably worked with StarOffice/OpenOffice and with Java, both with existing licences. And it is possible that the CDDL helps get around sudden (copyright and patent) licensing surprises and the need for binary blobs. But GPL compatibility is a pretty good measure of how reusable code is likely to be in other Free Software projects and is a desirable property even when dealing only with permissive licences. When something is not GPL- compatible, one wonders what kind of weird clause is in the licence and how it might affect further development and distribution.
Myself, I am a big supporter of copyleft licences, but I don't need someone in a corporation to legitimise my own preferences, particularly if they tell everybody else to do one thing while doing something different themselves (producing and advocating the use of proprietary software).
Paul
Hugo Roy hugo@fsfe.org wrote:
↪ 2015-11-01 Sun 12:20, Fabian Keil freebsd-listen@fabiankeil.de:
Mr. Fink also seems to knowingly conflate the Linux kernel with the free software community in general, for example when he spreads the FUD about DTrace and ZFS which according to him can't be used by "the community".
In that case, it's not far fetched considering the licenses for ZFS was apparently designed with the intention to make it incompatible with major existing licenses used in large shares by the free software community. (That's what I have heard repeatedly anyway.)
While I've heard that story repeatedly as well, another version is:
| The CDDL license, a file-level "copyleft" that allows open | source code to reside alongside proprietary code, was devised | by Sun in order to allow community development of the Solaris | kernel without having to negotiate the minefield of acquiring | rights to open-source code the company had licensed from | others, including IBM and NCR. In theory, any file with the | CDDL header in it is fair game for developers to build from. http://arstechnica.com/business/2011/12/disgruntled-employee-oracle-doesnt-s...
Free software licenses should reduce friction to allow greater flow of code, instead of deliberately increasing it.
That, and without forgetting that there are (were?) patents on ZFS stuff, which you cannot completely ignore (even outside the Linux kernel community).
The CDDL contains a patent grant so at least the patent claims from code contributors shouldn't be a big risk (assuming the patents are valid in the fist place).
AFAIK, the patent violations alleged by NetApp were never proven to be justified (and doubted by many). Unfortunately the details of the settlement haven't been leaked yet and the people involved seem to prefer not to talk about it in public.
While I agree that patents are a problem for free software in general, it's not obvious to me that using ZFS is more risky than using the Linux kernel or any other large code base.
In fact, I'd expect the Linux kernel to "violate" more patent claims than ZFS as its code base is quite a bit larger.
While the fact that they are licensed under the CDDL might be inconvenient for GNU/Linux distributions, for various other operating systems it's not a big deal and (from their point of view) even preferably to a more restrictive license like the GPL.
The fact that there are more permissive licenses than the GPL doesn't mean it's not a big deal to make a GPL incompatible license.
I never said that. My point is that other parts of the free software community are happily using ZFS and DTrace right now. Thus Mr. Fink's claim that the "the community" can't use them due to their license is ridiculous.
It's like claiming that "the community" can't use gcc versions after 4.2 because parts of the community have a policy to tolerate GPLv2 code in the base system but reject GPLv3 code due to the perceived risks for downstream consumers.
Fabian
Hi Fabian,
I mainly agree with you except on this one small point.
↪ 2015-11-02 Mon 13:57, Fabian Keil freebsd-listen@fabiankeil.de:
While the fact that they are licensed under the CDDL might be inconvenient for GNU/Linux distributions, for various other operating systems it's not a big deal and (from their point of view) even preferably to a more restrictive license like the GPL.
The fact that there are more permissive licenses than the GPL doesn't mean it's not a big deal to make a GPL incompatible license.
I never said that. My point is that other parts of the free software community are happily using ZFS and DTrace right now. Thus Mr. Fink's claim that the "the community" can't use them due to their license is ridiculous.
It's like claiming that "the community" can't use gcc versions after 4.2 because parts of the community have a policy to tolerate GPLv2 code in the base system but reject GPLv3 code due to the perceived risks for downstream consumers.
It's not the same claim. In one instance, there is a *choice* to reject the license (GPLv3) while, in the other instance, you *can't* use the license (CDDL) because it conflicts with the already-used license.
If you combine this with the fact that the CDDL itself has allegedly been chosen *because* (or designed so that) it's incompatible with a license that's used *a lot* by the community, I think it's really not far fetched to say the tactic was entirely anti-community and, instead, was pursuing another agenda.
This is also supported by the fact that they filed for patents on the technology, thus even making sure other implementations under a different license are excluded (which isn't the case with gcc, you can fork, reimplement, start a competing project, etc.).
The fact that some projects in the community can still use ZFS with the CDDL is just a minor side-effect.
Thus I disagree with the claim being “ridiculous” as you wrote and I wouldn't downplay what looks in retrospect as a very nasty tactic (which failed).
Best, Hugo
PS: please note that I used the word "allegedly" and "looks" -- I have no specific knowledge of what really happend on ZFS and behind the choice or design of the license. But there sure are some elements raising questions... and in the end we have a missed opportunity, because I really do hear that ZFS was awesome.
Hugo Roy hugo@fsfe.org wrote:
↪ 2015-11-02 Mon 13:57, Fabian Keil freebsd-listen@fabiankeil.de:
While the fact that they are licensed under the CDDL might be inconvenient for GNU/Linux distributions, for various other operating systems it's not a big deal and (from their point of view) even preferably to a more restrictive license like the GPL.
The fact that there are more permissive licenses than the GPL doesn't mean it's not a big deal to make a GPL incompatible license.
I never said that. My point is that other parts of the free software community are happily using ZFS and DTrace right now. Thus Mr. Fink's claim that the "the community" can't use them due to their license is ridiculous.
It's like claiming that "the community" can't use gcc versions after 4.2 because parts of the community have a policy to tolerate GPLv2 code in the base system but reject GPLv3 code due to the perceived risks for downstream consumers.
It's not the same claim. In one instance, there is a *choice* to reject the license (GPLv3) while, in the other instance, you *can't* use the license (CDDL) because it conflicts with the already-used license.
Nobody is forcing anyone to use the Linux kernel, thus there's at least one choice too (use another kernel).
Debian GNU/Linux users could, for example, migrate to Debian GNU/kFreeBSD. Obviously such a migration has downsides but it certainly is an option.
Another option would be to come to the conclusion that ZFS can actually be used with Linux under certain circumstances: http://zfsonlinux.org/faq.html#WhatAboutTheLicensingIssue
A couple of GNU/Linux distributions are already including ZFS: http://www.open-zfs.org/wiki/Distributions#Linux
If you combine this with the fact that the CDDL itself has allegedly been chosen *because* (or designed so that) it's incompatible with a license that's used *a lot* by the community, I think it's really not far fetched to say the tactic was entirely anti-community and, instead, was pursuing another agenda.
This is also supported by the fact that they filed for patents on the technology, thus even making sure other implementations under a different license are excluded (which isn't the case with gcc, you can fork, reimplement, start a competing project, etc.).
My impression is that lots of companies file patents to "prove" to their share holders how innovative they are and to protect against patent threats from competitors (that aren't patent trolls).
Former Sun Microsystems CEO Jonathan Schwartz has someone anecdotes about the latter case:
| I feel for Google – Steve Jobs threatened to sue me, too. | | In 2003, after I unveiled a prototype Linux desktop called | Project Looking Glass*, Steve called my office to let me know | the graphical effects were “stepping all over Apple’s IP.” (IP | = Intellectual Property = patents, trademarks and copyrights.) | If we moved forward to commercialize it, “I’ll just sue you.” | | My response was simple. “Steve, I was just watching your last | presentation, and Keynote looks identical to Concurrence – do | you own that IP?” Concurrence was a presentation product built | by Lighthouse Design, a company I’d help to found and which Sun | acquired in 1996. Lighthouse built applications for NeXTSTEP, | the Unix based operating system whose core would become the | foundation for all Mac products after Apple acquired NeXT in | 1996. Steve had used Concurrence for years, and as Apple built | their own presentation tool, it was obvious where they’d found | inspiration. “And last I checked, MacOS is now built on Unix. I | think Sun has a few OS patents, too.” Steve was silent. | | And that was the last I heard on the topic. https://jonathanischwartz.wordpress.com/2010/03/09/good-artists-copy-great-a...
Microsoft threatening OpenOffice is another example in the post.
The fact that some projects in the community can still use ZFS with the CDDL is just a minor side-effect.
Thus I disagree with the claim being “ridiculous” as you wrote and I wouldn't downplay what looks in retrospect as a very nasty tactic (which failed).
I may have missed it, but IIRC Mr. Fink merely claimed that "the community" can't DTrace on ZFS due to the CDDL but did not suggest that this was the result of a "very nasty tactic".
I consider the first statement to be obviously false and thus ridiculous while the second statement is hard to confirm or disprove from the outside and thus a valid opinion.
PS: please note that I used the word "allegedly" and "looks" -- I have no specific knowledge of what really happend on ZFS and behind the choice or design of the license. But there sure are some elements raising questions... and in the end we have a missed opportunity, because I really do hear that ZFS was awesome.
I agree that ZFS indeed was awesome (I've been using it since 2007). The good news is that it is still available and has become even better since then. For details see: http://www.open-zfs.org/
Fabian