this is a different SCO [LWN.net]

this is a different SCO

Posted May 20, 2005 17:59 UTC (Fri) by dmaxwell (guest, #14010)
In reply to: this is a different SCO by paulj
Parent article: Sun Seizes Tarantella (IT-Director)

There is still an implied patent threat against anybody who works on non-Sun open source projects. What's even worse is that yet again Sun is using a bizarro license instead of one of the established licenses like BSD, MIT, GPL and so forth. Which means yet again that improvements will flow into Sun's little open source island but none will get out. It is a net negative. BSD and Linux people will have go out of their way not to see that code.

Add all of this to your upper managment's ever changing position on Linux and FOSS and new found respect for MS and I have to wonder if someday Sun is just going to be Caldera on a larger and more dangerous scale.

Let's see binding promises not to threaten FOSS projects in any way and perhaps we can change our minds about Sun giving millions to SCO.

to post comments

this is a different SCO

Posted May 21, 2005 2:27 UTC (Sat) by paulj (subscriber, #341) [Link] (14 responses)

There is still an implied patent threat against anybody who works on non-Sun open source projects.

Well, I don't see why you would think that. Sun have no history of filing patent infringement claims, rather they've been on the receiving end of them (Kodak most recently, and there's an anecdotal (and possibly apocryphal - i have no idea if it's true) story of how IBM tried to shake them down for patent royalties early in their history; when Sun engineers destroyed the patent claim point by point, IBM's lawyers response was along the lines of "Do we have to go back to Armonk and find which of our other 20k patents you might be infringing on?"). Several Sun execs are on record, most particularly Jonathan Schwartz, stating that Sun has no intention of using patents other than defensively.

So I don't quite understand this comment. Is there a patent risk to Free and/or Open Source software? Sure there is, just as (unfortunately, in my personal opinion) in developing any software. Is there a risk Sun is going to go mad and launch patent litigation against Free Software? I doubt that very much, there's just nothing in Sun's corporate history and outlook to suggest that Sun would. IBM OTOH do have some history of litigation, but oh wait, they donated all those useful patents to open-source, and promised not to sue anyone involved in the Linux kernel (not free or open-source software generally, please note). Further, Sun sell Linux (eg JDS) and ship lots of open-source software (GNOME, OpenOffice, Samba, Apache, Webmin) - not only ship, but contribute to as well. Sun are quite pro Free/Open Source Software, and have a strong interest in not damaging the ecosystem.

What's even worse is that yet again Sun is using a bizarro license instead of one of the established licenses like BSD, MIT, GPL and so forth.

Firstly, Sun release software and contribute to projects under lots of different licences - as appropriate. Sun LGPL'd OpenOffice. They employ GNOME hackers to contribute to both GPL and LGPL GNOME projects. I contribute regularly to a GPL project. Sun contribute to Xorg -> MIT licence. Licencing is very much a pragmatic thing. (And, btw, I consider the GPL a very pragmatic licence :) ).

Secondly, the GPL was very seriously considered as the licence for OpenSolaris, I believe. But as you point out above, software development these days carries patent risks (Kodak, etc.). The Free Software community knows this too, indeed one of the key questions for drafting GPLv3 is how to tackle patent risk. The GPL, for pragmatic reasons, simply couldn't work for OpenSolaris. 1. Mixing proprietary modules with OpenSolaris would become potentially a grey area, and Sun wanted to be able to allow ISV's to provide binary modules (BSD spirit I guess, and it makes business sense), eg because the ISV either can not or doesn't want to ship source. That's pretty similar to how people are using binary kernel graphics/DRI drivers on Linux today, except the binary module vendors potentially are, and the 'repackager' sites that distribute these as RPMs or deb's even more so, are in a grey area wrt GPL. Sun preferred to avoid the grey area and use an MPL licence. As the MPL obviously needed to be modified (it's quite a netscape specific licence), Sun decided to tackle the patent litigation problem. Quite nicely imho, hopefully the GPLv3 will use a similar trick. They also spent a good bit of lawyer time on making the CDDL as generic as possible.

Which means yet again that improvements will flow into Sun's little open source island but none will get out. It is a net negative.

No offence, but I suspect you've bought the FUD without actually reading the CDDL. It isn't Sun's open-source island, if you think it is, I hope you're not reading this with an AOL-open-source-island Mozilla derived browser :). It's open-source full stop. Sun will have no power to 'take back' code licenced under the CDDL, Sun will have no power to stop people dual-licencing code under CDDL and whatever else. Also, code released under the CDDL can be mixed quite freely with code under other licences, the CDDL doesnt stop that. (The GPL OTOH /does/ prevent a lot of mixing, that's just the way it is.). Code released under an open licence is not a net negative.

BSD and Linux people will have go out of their way not to see that code.

Why? You've gotten this view, I suspect, from posts by a certain gcc/glibc hacker (who happens to work for a certain competitor of Sun - bear that in mind ;) ). There's just no basis for this opinion. Firstly, you can't infringe copyright by looking, looking is fine as long you don't copy. Though, obviously, it is easier to demonstrate you didn't copy if you can show you never looked, but looking and reimplementing is not a copyright infringement. There is no evidence that OpenSolaris and the CDDL are some kind of evil-genius venus-fly-trap designed to catch unsuspecting Linux hackers. I personally (very much a personal opinion) find the idea that Sun would sue anyone for looking at CDDL'd Solaris code and reimplementing it in Linux or BSD or wherever absurd. There are already plenty of reimplementations of Solaris features in Linux and glibc (slab cache allocator, NFS, RPC) and I note a distinct lack of lawsuits..

Secondly, if it is patent infringement you're worried about, well not having looked doesn't protect against patent infringement suits, indeed not having known doesn't protect you. Whether you have or have not looked is simply irrelevant to patent infringement. And again, Sun sueing open-source for patent infringement would be very strange (Sun ship lots of it, including Linux..).

new found respect for MS

Well, MS violated a licence on Java. After years of trying to gain judicial redress, Sun got a settlement out of it. If someone does significant wrong to you, there's nothing bad in ultimately being compensated for that wrong. Part of that settlement included MS cooperating with Sun on interoperability issues. If you have a mixed Solaris and Windows environment (and how many companies don't have Windows machines in their server room these days?) this is a good thing. Executing on better interoperability requires some kind of decent working relationship.

Let's see binding promises not to threaten FOSS projects in any way and perhaps we can change our minds about Sun giving millions to SCO.

I wish I were a Sun exec and I could give it to you in writing. I'm not, and I can't however. I can though point you to the text of the CDDL which requires all contributors to grant all patent rights relevant to contributed code to all developers/users of that CDDL licenced code. But I guess that isn't enough, as it doesn't match the far better promises made by all those other "good" FOSS companies. IBM's commitment to not sue the Linux kernel for violating patents on tamper-proof screws or gel-packs: aye, granting rights to all your relevant patents to OpenSolaris and any future derived CDDL projects just doesnt compare.. ;)

And the SCO millions, well that's pretty much what it cost to ensure Solaris could be open-sourced, according to Scott McNealy. The timing was unfortunate. That's simply the cost of GNU/OpenSolaris.

I know it's quite chique to be anti-Sun, or least Sun-sceptical, but if you look past the FUD, and actually look at what Sun have done and do wrt FOSS, and what Sun execs actually say, you might find there isn't much basis to a lot of it. You don't have to like Sun - yeah, Sun are very commited to Solaris, Sun have spent a lot of time on it, and Sun don't intend to replace Solaris with Linux+glibc, for good reasons. Strangely, the BSD folk, who are as proud of their work as Sun are of theirs, dont seem to want to switch kernels+core userspace either - but at least be fair to yourself in how you assess these 'fashionable' opinions.

Also, bear in mind that one of the key strengths of Unix/Linux is the healthy competition of multiple implementations. There's pretty much only *BSD, Linux and Solaris left now as viable, healthy Unix implementations. All of them are (or will be by next month) Open Source / Free Software. The Unix ecosystem is enriched by this diversity and competition, for that reason alone Solaris is valuable (even if you don't use it. Eg, I don't use BSD, but I very much appreciate their existence.).

--paulj

(Paul Jakma, FSF 'member', Sun employee, Free Software contributor, as are many of his colleagues. Opinions above are very much my own, and based on information stated in public, eg in the press or blogs.sun.com.)

Paulj, Paulj

Posted May 21, 2005 6:48 UTC (Sat) by Tao (guest, #24985) [Link]

I am amazed that you bothered to respond to those cheap shots and nonsense. I feel your pain, and your passion too.

Regards,
Tao
(no, I don't work for Sun.)

The problem is clarity

Posted May 21, 2005 14:52 UTC (Sat) by dmaxwell (guest, #14010) [Link] (12 responses)

Early in the process, Sun was asked what their reaction would be to code in other projects that seemed to embody those patents. The answer was along the lines of "we reserve to right to utilize our intellectual property". It is very hard to take that as anything but a threat UNLESS it is Sun's open source that is being contributed to. They could have allayed that worry very easily; they chose to play it up.

Realistically, there are numerous defenses against a Sun that is going litigation crazy. You rightly point out they have been involved in FOSS on numerous levels. Nonetheless, Sun has had a hard time these last four years. And I really don't think you can deny the ambivalence Sun has about Linux and really FOSS in general. They're for it when they think it will be helpful and dead set against it when they think it won't. There seems to be no consistent underlying philosophy about when it is one or the other. I don't know what to think about Sun's positions and intentions because the story about what just they are continually changes. You say that Sun isn't evil right now; I actually buy that. You also say they'll never be evil; that could change next week....literally. The best and easiest thing Sun's upper management could do for their image is to pick a story and stick to it. At least we could know what to think instead of necessity forcing us to assume the worst.

We've all heard the "decapitated Tux" story.(perfect illustration of my point BTW. Wear a Tux suit to show that Sun too can provide Linux solutions. Then months later, publically talk about stuffed and mounted penguins...) It isn't hard to imagine scenarios in which a Sun in dire straits turns to the lawsuit lottery to stay afloat. The SCO example is no deterrent because everyone knows they are a paper tiger. Sun isn't.

The problem is clarity

Posted May 21, 2005 18:19 UTC (Sat) by paulj (subscriber, #341) [Link] (11 responses)

Hmm, I don't remember that. Further, that's a nasty and loaded question to be asked. Anyone with corporate responsibility who's asked "Will you indemnify a huge and vaguely defined swathe of software developers against infringement of your patents?" really has no alternative but to answer in the negative, "We reserve the right" is as good an answer as could be given, and is not at all a threat. Any other answer would have been irresponsible for a corporate officer to give. Hence the accusation of "playing up the patent threat" seems quite unjustified.

You simply can not give an off-the-cuff answer here of "sure why not", as such a remark would literally have significant legal standing. It's a matter of a core corporate position which would need to be discussed and approved internally first. And guess what, Sun have since done that and the result is a unilateral patent right grant of all Sun's patents which are relevant to any code released under the CDDL, which, by recent public account, should include all of the core OS/Net portion of Solaris, and more, by next month[2]. (And the CDDL is a fairly liberal licence, so that means both code implementing those patents and the patent grant will be available to lots of Open Source code. Even GPL code, if the GPL copyright holder gives an exception to allow linkage).

But I guess far easier to take "We reserve our rights" and spin that into "They're evil!" than to use one faculties to examine all the information and opinions with a critical eye (including the FUD) and see whether it stands up. (No offence intended, I simply feel a lot of the FUD wouldn't stand up if one were to examine it more closely).

Nonetheless, Sun has had a hard time these last four years.

Yes, true. AIUI, it's not the first time Sun have had to reexamine their strategy. The result of has been the move to open-source Solaris, which hopefully is to come to fruition Real Soon Now, after literally years of work. If anything were a sign that Sun has decided to embrace open-source, then that should be it.

That's exciting, even if you don't use or care for Solaris, simply because it is generally good for Unix (in the generic sense, ie including Linux). Every Unix implementation benefits from the competition which is provided by other healthy implementations. Every Unix you look at, you'll probably find features there which came from some other. Not to mention the huge body of general software which is portable between them. And quite a lot of now common Unix features were originated by Sun (I forgot things like PAM and NSS from my previous list, and there are surely more ;) ).

And I really don't think you can deny the ambivalence Sun has about Linux and really FOSS in general.

FOSS in general: Sorry, Sun are quite supportive of FOSS imho. Linux: Well Sun think Solaris is better, what would you expect :), other than that I dont think Sun have a problem with Linux. Commercial competition from Linux-pushing companies like RedHat and IBM? Never mind ambivalent, Sun very much intend to battle as hard as it takes! :).

The ambivalence possibly arises from confusion on Sun's position between these various strands (FOSS, Linux, commercial and pure business competition). And yes, some of that confusion likely arose out of early confusion by Sun in distinguishing between those 3 seperate things. I think that confusion is gone now though.

FOSS: Good!
Linux: Solaris better :)
RedHat/IBM/et al: the competition!

Clearer? :)

The best and easiest thing Sun's upper management could do for their image is to pick a story and stick to it. At least we could know what to think instead of necessity forcing us to assume the worst.

I agree. I think Sun management have realised this already. Hopefully the 'message' from Sun exec's the last while has been clearer, and will remain so[1].

It isn't hard to imagine scenarios in which a Sun in dire straits turns to the lawsuit lottery to stay afloat.

That seems quite an unlikely scenario to me.

1. Note that I've never before been employed by a company where I feel I could be that candid in public about my boss's boss's boss's boss's :).

The problem is clarity

Posted May 22, 2005 11:19 UTC (Sun) by hansl (subscriber, #5086) [Link] (10 responses)

> Linux: Solaris better :)

How? I mean, as far back as in 1998 me and my fellow sysadmins
were asked by our management to make a choice. At that time we
were running a mix of Irix, SunOS and Linux and we had to renew
support contracts as well as buy some new hardware.

Management didn't want three support contracts for three different
platforms so we were asked to pick one platform. We chose Linux
(Redhat) because it fullfilled all of our needs and it ran on fast
and cheap PC hardware (so we and the Windows group could share a
standardised hardware platform).

Redhat's support is simply amazing and doesn't cost you anything
extra either. I created several bugzilla reports and at first I was
shocked that they even responded. Later I got used to hearing: "been
fixed already".

I fail to see how Solaris with Sun's expensive support on expensive
and comparatively slow hardware can ever catch up with this.

You see, I never mentioned the words "license", or "open source"
until now. Even from an economical viewpoint Solaris doesn't make
sense to me.

But still, it's the open source *process* I trust, not the company.
We could go with Novell/SUSE if Redhat fails to deliver. Even with
Solaris aspiring to take part in the open source process I don't see
how I could switch to another company to get support for it.

-Hans

P.S. I don't work for Redhat.

The problem is clarity

Posted May 22, 2005 13:24 UTC (Sun) by dmaxwell (guest, #14010) [Link] (1 responses)

As far as Sun's "slow hardware" goes, it seems the x86 ports are getting more attentions. And Sun seems to be pushing branded Opteron boxes. I would think that takes care of much of the slowness and expense associated with Sun hardware. Whether or not those Opteron boxes will have the bulletproofness of classic Sun boxes remains to be seen.

As for the "process", I'll wait and see on that as well. Sun's engineers will be as good as ever. However, I suspect the custom licensing will limit the size of the development community Sun hopes to create.

The problem is clarity

Posted May 22, 2005 14:07 UTC (Sun) by hansl (subscriber, #5086) [Link]

> Sun seems to be pushing branded Opteron boxes

How is that going going to help? I bet those "branded Opteron
boxes" are going to be more expense than non-branded Opteron
boxes that Redhat is willing to give me support on.

And how can Sun afford to compete with it's own multi-billion
dollar hardware investment called Sparc? By letting me pay for
it? No thanks.

As for the bulletproofness of Sun hardware, you can buy 2
Opteron boxes for the price of one Sun box to achieve higher
availability as well as more than twice the performance of
that single Sun box.

-Hans

The problem is clarity

Posted May 23, 2005 1:14 UTC (Mon) by paulj (subscriber, #341) [Link] (7 responses)

Redhat's support is simply amazing and doesn't cost you anything extra either.

I don't know anything about RedHat support, no experience of it, but last I checked it started at $179 for RedHat Workstation Basic support (30 days support on basic install/configure, 1 day response) up to $2499 for RHAS premium edition. That doesn't quite seem to square up with "doesn't cost you anything". Sun's Solaris 10 support offerings OTOH seem much cheaper. Eg $120 for Basic edition - and that's good for a whole year, unlike the 30 days of RHEL WS - to $360/socket for Premium edition, which is cheaper by a grand than RHEL AS Premium on a four socket machine. (Eg, a 4 socket Sun v40z Opteron box with multicore chips, 8 CPUs, $1440).

And of course, the support offerings are optional with Solaris 10. You don't have to agree to a contract to pay Sun for support and maintain books on how many copies you've installed, and allow Sun to audit you, if you don't want to - unlike with RHEL and RedHat. No need to pay the RHEL tax, or any other OS tax, if you want to run Oracle on an Oracle supported OS. Just download Solaris 10 for free.

Note that I'm not in support so I'm not really at all au-fait with Sun's support offerings. The above information is just what was immediately accessible on www.sun.com. Sun seem to beat RedHat on price on all the offerings AFAICT, and Sun certainly are very keen on ensuring that that be the case. So that shouldn't be a good tack to take to argue against Sun.

As for slow hardware, sorry but 4 socket Opteron's really do scream, even more so when populated with aforementioned multicore Opteron chips :). (And yeah, you can buy them with Linux.)

Anyway, I'm starting to sound like a marketing droid, which worries me.

The problem is clarity

Posted May 23, 2005 7:48 UTC (Mon) by hansl (subscriber, #5086) [Link] (6 responses)

> up to $2499 for RHAS premium edition. That doesn't quite seem
> to square up with "doesn't cost you anything".

I'm not talking about that type of support. We are three RHCE's
here, we can very well do without "installation" support. No, I
meant that I can file a bug report or request an enhancement
in Redhat's bugzilla without having to pay for it. Which makes
sense because I'm helping them to fix a problem I have with
*their* software.

Don't know Sun's prices but if you think about it's ridiculous
that you have to pay on the order of $20,000 to be able to help
Microsoft fix a problem with their software, isn't it?

-Hans

The problem is clarity

Posted May 23, 2005 8:11 UTC (Mon) by paulj (subscriber, #341) [Link] (5 responses)

I'm not quite sure, but I think that has been changed.

--paulj

The problem is clarity

Posted May 23, 2005 9:28 UTC (Mon) by hansl (subscriber, #5086) [Link] (4 responses)

> I'm not quite sure, but I think that has been changed.

It better have. Because Sun isn't going to benefit from the
open source process just by putting (parts of!) it's source code
under an open source license and then charge for services and
support.

And BTW, they made the wrong decision at that IMO, it should
have been a reciprocal license like the GPL because that's the
type of license that creates more trust than any other open source
license.

To really enjoy the benefits of open source Sun should open up
their development process as well. For all the brilliant Sun
engineers cannot do what the community seems to be doing for
Linux.

A good bug report, a patch that adds a PCI id for some obscure
hardware or a one-liner bugfix is just as valuable as the
next hyperthreading aware scheduler by some brilliant engineer.

Because for the end-user, all of these can equally make a piece
of hardware or software start to work.

-Hans

The problem is clarity

Posted May 23, 2005 10:14 UTC (Mon) by paulj (subscriber, #341) [Link] (3 responses)

Other than the licence comment, I agree with everything you said, and I believe these things (bug reports, transparent development, etc) are possibly being addressed for the preparation of OpenSolaris, so wait and see until that goes live.

Re the licence, it *is* a reciprocal licence. Code under CDDL has to stay under the CDDL, along with derived works. The major differences with the GPL are that CDDL does not try to extend its reach beyond *files* containing CDDLed code (In this sense it less restrictive than the GPL, and while I'm sure RMS would frown on this, it's a more pragmatic approach which suits OpenSolaris better, see my previous comments on this earlier in this thread. And to para-quote Linus "He who writes the code gets to choose the licence".) and it attempts to deal with patent risk, quite nicely too imho and I hope the GPLv3 will do take a similar tack. I don't see any barrier in the CDDL to creating a healthy community around a body of source-code.

If I thought the CDDL was a terrible licence, I'd just shut up and say nothing. But it's not a bad licence to my reading (ie reading the actual licence - not from reading the inane commentary on /. and elsewhere from people who seem *not* to have read the CDDL at all), and I say that with my FSF member hat on :).

regards,

Paul Jakma.
(Who has been paid to work on Free Software for over a year now by Sun, and long may it continue, fingers crossed.)

The problem is clarity

Posted May 23, 2005 11:02 UTC (Mon) by hansl (subscriber, #5086) [Link] (2 responses)

> The major differences with the GPL are that CDDL does not
> try to extend its reach beyond *files* containing CDDLed code

So people only have to make sure their changes are in separate
in files and they don't have show them to me (when distributing
the aggregate work).

That's not the reciprocal "you get to use my changes, so I get
to use your changes" deal that the GPL establishes, which is
exactly why it becomes interesting for me to contribute.

-Hans

The problem is clarity

Posted May 23, 2005 11:39 UTC (Mon) by paulj (subscriber, #341) [Link] (1 responses)

Well, why not read the CDDL instead of getting a second-hand opinion from some random, non-lawyer, LWN reader like me? (this is precisely my problem with so much of the opinion bandied around on the CDDL - everyone is happy to form it based on hearsay. :) ).

You can't simply make changes and stuff them in a new file. Modifications to CDDL'd code must still be distributed under the CDDL, whether in a new file or not. The definition of modifications is:

---------------------------------------------------------------------
1.9. Modifications means the Source Code and Executable form of any of the following:

A. Any file that results from an addition to, deletion from or modification of the contents of a file containing Original Software or previous Modifications;

B. Any new file that contains any part of the Original Software or previous Modification; or

C. Any new file that is contributed or otherwise made available under the terms of this License.
-----------------------------------------------------------------------

Such modifications would *still* be required to be licenced under the CDDL.

In that context, reread this sentence fragment of mine:

> The major differences with the GPL are that CDDL does not
> try to extend its reach beyond *files* containing CDDLed code

Clearer?

Note that the definition of "modification of the contents of a file" *possibly* is very similar to how the GPL relies on the definition of "derived work" for its coverage. But I'm not a lawyer, so I don't know. (I'd be curious as to the answer). The scope of how the CDDL frames it is possibly narrower than that of the GPL, I'm not sure (modifications to contents versus 'derived work') however I suspect in practice the respective scope wouldn't be /too/ far apart.

Reading the CDDL would also be a useful exercise, particularly if one wishes to argue it's demerits :). It's at:

http://www.sun.com/cddl/cddl.html

regards,

--paulj

The problem is clarity

Posted May 23, 2005 12:53 UTC (Mon) by hansl (subscriber, #5086) [Link]

> Well, why not read the CDDL instead of getting a second-hand
> opinion from some random, non-lawyer, LWN reader like me?

OK, I actually started reading the license but I'm afraid
I am not skilled enough to understand how exactly it allows
you to not reveal your own changes. But I think I do understand
these words from your boss Jonathan Swartz, explaining the CDDL:

> Under the CDDL, you are free to choose what to reveal, what to
> withhold, and how to price your products. The CDDL encourages
> self-determination by giving developers the basic building blocks
> of the entire OpenSolaris operating system without any obligation
> to disgorge their private property or predetermine its price. You
> can withhold from Sun, from the OpenSolaris community, from the
> world, anything you build. Or you can choose to share. It's entirely
> your call.

http://blogs.sun.com/roller/page/jonathan/20050417#the_sp...

-Hans