OSNews

I was just about to say the same thing.

"It's a great shame that OpenSolaris still exists. They should have killed it," said Morton

What! Isn't that like the exact opposite of the whole free software movement?! The competition should kill themselves and join the rest of us??

I'd love to hear what Morton has to say about KDE and Gnome!!!

The trouble seems to be related to the number of people in the free software/OSS/whatever movement who take a decidely-fundamentalist stance. Not in any sort of religious sense, before anyone goes into "rrrawr, seeing red!" mode - but in the sense of an unwavering conviction that a particular ideology is the only possible "right" one, and the world would be a wonderful paradise *if only* everyone else would just accept the inherent "rightness" of their way of thinking.

The issue at bar is ultimately license fragmentation. Copyleft is a good idea, but it stops being practical when everybody and their dog creates their own copyleft license. The GPL came first, and Linux is, for better or worse, licensed under the GPL.

OpenSolaris is the first kernel to use a non-GPL copyleft license. We've never had competing copylefts at the kernel level until Sun made it so. Developing under copyleft is like blowing a bubble. Everybody developing under the GPL is blowing into the same bubble.

Now Sun has started blowing their own bubble. Developers have to choose which bubble to blow on. There's no way that their breath can inflate both bubble simultaneously. The bubbles have to grow independently, even if everybody would like to see them merge.

Competition is good. Fragmentation is bad. If Sun had decided to license OpenSolaris as GPLv2 or later, that would have been a great day for free software. Competition within the bubble, where developers can compile the best work from either project, would be outstanding. That's the great shame to which Andrew refers.

Now, it's not really Sun's fault. Copyleft licenses are unabashedly egocentric. They're a model of universal gravitation for software. Code falls into orbit around its copyleft license. Following the analogy, each copyleft license is a distinct solar system. Before we had the GPL solar system plus some comets of the BSD persuasion. Now we have two solar systems.

So while it's easy to see the Linux developers as close-minded or fundamentalists, they're really just slaves to the license. Copylefts don't coexist. They're isolationist and protectionist. Like it or not, this is an implication of copyleft. Again, it's a great shame.

As much as I admire Solaris on its technical merits, it's existence is unfortunate for Linux. Likewise, the existence of Linux has been unfortunate for Solaris for years. Copyleft is a join-em or fight-em proposition. Whether or not this is good for users is highly debatable.

Microsoft isn't splitting the Linux community with patent covenants nearly as much as Sun is splitting the free software community with the CDDL.

but but but... i thought they were going to make solaris gpl v3?

That would still be incompatible with the GPLv2 as used in the Linux kernel. Unfortunately, even if Sun and the Linux kernel community sit down and talk about license compatibility, there's no practical way for Linux to meet Sun halfway. Linux is pretty much trapped by its v2-only license and distributed copyright ownership.

The least invasive way to mitigate this license dilemma is to have an arrangement where Sun will consider relicensing specific portions of the OpenSolaris under GPLv2 or later upon the request of the Linux community. As it stands there is nothing preventing Sun from integrating Linux code under the GPLv2 with OpenSolaris under the CDDL.

That would change if OpenSolaris goes GPLv3. Then Sun would have to (at least) request that the copyright owners of portions of the Linux kernel relicense their code as GPLv2 or later.

I find it hard to believe that Sun will add a GPLv3 option for OpenSolaris. Its existence imposes restrictions on Sun integrating code from non-GPL projects. Furthermore, as used in a dual-license setup with the CDDL, there is no practical reason for recipients to elect to abide by the more restrictive GPLv3.

However, there's a distinct possibility that Sun could drop CDDL in favor of LGPLv3. They're very similar licenses, but the LGPLv3 would cut down on license proliferation and give Sun a boost in street cred. The primary reason why Sun did not use the LGPLv2 (or to a lesser extent the MPL) is because it lacks strong patent grant and non-aggression clauses.

Heck, Sun could choose to kick CDDL in favor of Ms-CL. That would be an interesting turn of events.

OpenSolaris is the first kernel to use a non-GPL copyleft license. We've never had competing copylefts at the kernel level until Sun made it so. Developing under copyleft is like blowing a bubble. Everybody developing under the GPL is blowing into the same bubble.

I think you should stop drinking FSF's koolaid. There have been other kernels that were released with non GPL licenses. BSD comes to mind.

When did BSD become copyleft ?

Regarding...

I agree with you. What an absurd thing to say. These projects foster creativity, competition and different minds to solve problems in different ways.



Don't be so sure Mr Morton. The past few years have seen Sun slowly evolve into a more open and transparent provider of solid, well known and supported platforms.

Just because Sun hasn't done it yet, doesn't mean it will not happen. Most recently some of their hardware specs have been made available under GPL, so your argument that they refuse to adopt the GPL is inaccurate at best.

There seems to be a change of attitude at Sun which is exciting and valuable to the community and to business.

Maybe the air was a little too thin up there on stage.

It reminds me of Linus Torvalds who asked every Linux user to move to KDE. Some GPL zealots are really stupid, they forget too fast the core philosophy of free software and having an open-minded vision to improve and to share software without any restriction.

It's not really the same. Torvalds said that because he thought GNOME was going down the wrong path *on technical merits* with respect to interface design (as it were, the dumbing down of the interface). It was hardly ideological, although the flamewar ended up going that way.

Some GPL zealots are really stupid, they forget too fast the core philosophy of free software and having an open-minded vision to improve and to share software without any restriction.

Sun is really the one forget the core philosophy of free software and having an open-minded vision to improve and to share software without any restriction, by creating an
license that is incompatible with the most widely use license the GPL.Sun wants everything for themselves. If Sun really wanted to share code with linux,they would not of created the license that opensolaris is released under and had adopted the GPL,etc. I do hope Sun does release opensolaris under the GPLv3.

Edited 2007-08-10 00:38

It had nothing to do with licenses. Linus Torvalds argued against using Gnome because of what he (correctly IMHO) saw as "dumbing down the interface" and because of (his words) the "interface nazis" controlling development.

It was a technical issue and never ideological.

Hmmm, specifically to OpenBSD, since its base system (4.4 BSD) exists before Linux, Linux itself shouldn't exist, at least in Morton's view.

I'm surprised too. Are we sure we can trust the report? Could this be a bad recapitulation of a more complex reasoning?

Edited 2007-08-09 17:52

I think he's right. The sun-license is stupid. They allow closed-source software to adapt their code (Apple) but not GPL software. It should rather be the other way.
Sure, competition is good, but they would still compete if they were both GPL, and they'd both be better.

I think he's right. The sun-license is stupid. They allow closed-source software to adapt their code (Apple) but not GPL software.

Let's think for a second. Sun's license is compatible with most licenses and many projects now use Sun's code. It only won't work with GPL and suddenly it's Sun's fault. Why doesn't Linux change it's license so that everyone can use its code not just GPL projects? Which is truly open?

Edited 2007-08-09 19:33

Because GPL is more about free than open maybe?

>Sun's license is compatible with most licenses

But given that the GPL has a majority of the projects now, it is incompatible with most of the projects.

I've not heard many people claiming that CDDL is better than GPL so they'll switch, it's just different and incompatible.

>It only won't work with GPL and suddenly it's Sun's fault.

Yes, CDDL was conceived when the GPL was already widely used, if GPL had been made after the CDDL was widely used then it'd be the FSF fault, this is not the case.

> Which is truly open?
The one who is compatible with the *existing projects*?
As you see, it's a stupid question: everyone is free to define what 'truly open' means.

It only won't work with GPL and suddenly it's Sun's fault. Why doesn't Linux change it's license so that everyone can use its code not just GPL projects?
GPLv3 have been designed work with the Sun License and many other (APL notably). So at least parts of "linux" is trying to open up and fix the problem from our side.

But its perfectly alright for GPL to suck in code without giving anything back. Look at how many bits of code have been sucked in and relicenced under GPL to stop sharing back to the original author.

You're attacking 'proprietary vendors' and yet the GPL is the equivalence of a proprietary licence which prohibits code sharing.

btw, CDDL code can be shared with BSD projects - *BSD and Solaris are sharing code; it seems to me that the GPL world are the odd ones out.

"The sun-license is stupid."

So what? Their software, their rules. If you dont wanna play by them don't use it.

Sun's license doesn't disallow integration with GPL software. The GPL does. (That's irony, isn't it? So much for freedom.)

I think this will be one of those classic "I-didn't-mean-it-like-this(-even-though-in-reality-I-do)" tales.

Well, there's always the not so small possibility that zdnet reporter took his word out of context..

That said, Sun is really being quite nasty when they say that they hope that DTrace will be integrated in the Linux kernel *and* using at the same time a GPL incompatible license.

If they 'genuinely wanted to have it integrated into Linux kernel'(their word not mine), they would have chosen a compatible license, so I can understand A. Morton being annoyed by Sun's double talk.

Just to be clear: I think that Sun has every right to choose a GPL-incompatible license if they wish to keep a competitive advantage over Linux, but in this case they should just admit it, not the "we hope that Linux will integrate our code" knowing full well that this is impossible (unless they change their license of course).

Oh, there is also a debate in FreeBSD over DTrace: some of the dev believe that the CDDL license apply also to the headers and prevent them to link to DTrace so they have chosen to not integrate DTrace in FreeBSDv7.
It's not sure that the concern is really valid or not, it's still debated AFAIK.

What competition, it's not like linux doesn't have any. To me OpenSolaris is a stab in the dark from Sun at opensource, just like Novell did with SUSE.

Linux killed Irix.
Linux killed SCO.
Linux killed Dec-Unix.

Well... (Open)Solaris is not yet killed.

Linux killed Irix.
Linux killed SCO.
Linux killed Dec-Unix.

Well... (Open)Solaris is not yet killed.

I disagree. Itanium killed Irix, the HP merger killed Tru64, and plain market forces killed SCO*. I think that all three were killed by the general movement of comoditization of platforms, which they resisted much harder than some competitors. All three were ailing long before Linux was taken seriously in enterprise IT, and all three were backed by bad business plans.

* Actually, SCO is not EOL'd as far as I know. But I doubt they sell to any new accounts.

Anyway, this Morton thing, if it is true, sounds like a classic case of throwing his toys out of the pram on account of someone having the temerity to compete with him.

tually, SCO is not EOL'd as far as I know. But I doubt they sell to any new accounts.

Just FYI, have you ever used UnixWare? I believe it is unsellable. $699 for one CPU. Uses X11R5 (no X11R6 or even X.org). It has almost zero hardware support on hardware newer than 5 years, outdated tools and so on.

Not when the competition sports features that Linux doesn't have and won't have for foreseeable time, apparently ....

This is sad since both projects have their own objectives... Morgan might as well start dumping on the idea of different distos since that might lead to confusion which in turn might make people choose something 'not Linux' for their particular use....

Andrew - grow up, will you?

Linux has these things already.

Zones... Linux has had UML for a LONG time which does the same thing (run multiple copies of the same system image), plus XEN, and KVM virtualization capabilites built in. Solaris is the one lacking here.

DTrace.. Linux has SystemTap, which is 90% of what DTrace is and very young and getting better fast

ZFS - big deal a filesystem plus a volume manager built in... welcome to the 90's solaris. Most of us have had REAL LVM's for a while now, while sun has made us suffer with the peice of junk SDM for a long time or forced us to purchase the expensive vxvm product for something that should have been built in from the start. ZFS still has many stability issues, cant be used for boot...Linux already has high quality filesystems, and a more than production/enterprise ready LVM built in....Again, its solaris that is playing catch up.

Linux has these things already.

(snip)

ZFS - big deal a filesystem plus a volume manager built in... welcome to the 90's solaris. Most of us have had REAL LVM's for a while now, while sun has made us suffer with the peice of junk SDM for a long time or forced us to purchase the expensive vxvm product for something that should have been built in from the start. ZFS still has many stability issues, cant be used for boot...Linux already has high quality filesystems, and a more than production/enterprise ready LVM built in....Again, its solaris that is playing catch up.

What don't you like about SVM? There is nothing particularly special about it, but I have always found that it does the job. Actually, I have always liked that Sun did not bundle Veritas FS or VM, because it provided choice. Most other UNIXs of the era (HP-UX for sure, IRIX and Tru64 I think) licensed expensive third party VM code which you had to pay for whether you needed it or not.

Anyway, as long as the Linux folks keep sticking to "We already have feature X, so shut up about it" while paying no attention when the competition is actually innovating, I think Sun has a chance of turning the tide. It's almost too late, but not quite.

> plus XEN, and KVM virtualization capabilites built in. Solaris is the one lacking here.

I think you are a little behind the times. A visit to www.opensolaris.org will help you bring yourself up to date

Linux has these things already.

No it doesn't.

DTrace.. Linux has SystemTap, which is 90% of what DTrace is and very young and getting better fast

SystemTap is not 90% of what DTrace is. DTrace can dynamically instrument the entire software stack, which includes languages such as Java. SystemTap can't. DTrace can trace during bootup - great for debugging drivers. SystemTap can't. DTrace is also safe for production use and has been used in production for over two years... For other details for comparison, google DTrace.

Where did your value of 90% come from?

SystemTap ... and very young and getting better fast

Firstly, "getting better fast" - I could comment, but it may be best to let the SystemTap engineers to speak for themselves (and please post some release dates while you are at it :-).

As for "very young"... Is this some romantic Linux fantasy, where the courageous Linux developer defeats the Mighty Evil Corporate by trumping their product in a shorter time frame? Well, SystemTap has already missed the boat for that one.

The real story goes like this,

DTrace began work in October 2001, and was integrated into Solaris in September 2003, with the first customer access a couple of months later. That's 24 months. As far as we know, SystemTap began work in January 2005. It is now August 2007, 32 months later, and it still under development.

DTrace was written by 3 kernel engineers. I don't know how many engineers work on SystemTap, but when Mike Mason (SystemTap developer) was asked this question at LinuxWorld yesterday - his answer was unclear, mentioning 4 or 5 at one company, 4 at another, and trailing off. It sounds like there are about 10.

Team DTrace delivered a product with fewer engineers in a shorter timeframe, and they invented it.

If you want to point out other cool things about SystemTap, that are actually factual, then great.

Zones... Linux has had UML for a LONG time which does the same thing (run multiple copies of the same system image), plus XEN, and KVM virtualization capabilites built in. Solaris is the one lacking here.

Sun has virtualization that supports advanced RAS features on thier smallest 2U systems the T2000 that Xen and Linux don't support.

http://blogs.sun.com/narayan/entry/ldoms_virtual_io_failover

Linux KVM and Xen's Dom0 have a single point of failure which brings all the virtual machines down. With LDoms the primary domain or I/O service domain can crash and the guest VMs will pause till the service comes back up or like the demo in the blog shows with IPMP and fail over you can have uinterrupted service.

That's the problem of some people in the FOSS world. Everything should be open according to them but they can't "open" their mind.

The same people are against Mono for example. They systematically refuse everything that didn't originate in the FOSS world. That's sad because there are many good things that come from the outside.

Anyway, I think it's all politics and business again. Andrew is paid by Google and Ian by Sun. And Google is all the way in the Linux camp.

I'm sure there are people at Google that are waiting for the right time to invest much more money into Linux to make it profitable for them.

You talk as if we all are demanding for a full GPLv3 kernel, and certaintly most of us are pretty happy with Linux and the GPLv2. Only the FSF is screaming for someting like that and the FSF != us or mayority.

GPLv3 zealots like to talk like GPLv2 licensed software were now "inmoral"......I'm sorry, GPLv2 licensed software is free software, In fact for many people (like linus) the GPLv2 is a better free software license than the less free GPLv3. Do you really think free software developers care about a 100% GPLv3 distro....?

Opensolaris taking the linux market share? Just because they change the license? Don't make me laugh. In the last years it has been pretty much the contrary - linux has stolen thousands of solaris installs (and not only because of the license)

Opensolaris can and will take a lot of market share from the remaining old, propietary UNIX users, BSD users, windows users....but linux? I don't think it'll steal more than a few of them. Seriously.

That's the most brillant comment I've ever read on OSNews.

I'm glad OpenSolaris exists.

Think about these feelings the next time you and your colleagues suggest that the Linux community just give up on SystemTap and port DTrace instead. There is undeniable pride on both sides of aisle. Nobody likes to read in the press that somebody wants their project to die.

Linux's rapidly-maturing tracing framework has just as much right to exist as Sun's rapidly-maturing free software community. I think that Andrew's comments are regrettable, but can you honestly argue that Sun doesn't have just as much NIH syndrome as Linux?

Let's agree to disagree. It doesn't appear like either side is willing to consider each other's work. So you guys go do your thing, and the Linux folks will do their's. We'll have competing solutions in several areas, and customers will be forced to decide which platform offers the better overall value for their needs.

Unfortunately, they won't get to mix and match. It would have been nice if Sun decided to throw their weight behind Linux. I think that together we could have developed a *nix platform of unprecedented strength. But in the end, Sun decided that they could do better on their own. So we can't be friends anymore, it seems. Please stop calling...

NOTE: I have nothing against OpenSolaris. I simply got into Linux while Solaris couldn't run on any of my hardware. I've grown fond of it. I don't think that the existence of OpenSolaris will detract from Linux development. So in the interest of furthering free software in general, I wish you guys the best of luck.

Think about these feelings the next time you and your colleagues suggest that the Linux community just give up on SystemTap and port DTrace instead. There is undeniable pride on both sides of aisle. Nobody likes to read in the press that somebody wants their project to die.

I think you're missing the difference between competition and, for lack of a better diagnosis, psychotic monomania. It's true that DTrace and SystemTap compete, just as OpenSolaris and Linux compete, but I have never said and would never say that it's a "shame" that either SystemTap or Linux exists, because it plainly isn't: I believe to my marrow that choice -- and therefore competition -- is fundamentally a Good Thing. I believe this in part because I've seen the work that it forces out of me personally: I did much better work when Linux was clobbering us than I did when we were gorging ourselves on the spoils of the Dot Com boom. So I welcome competition, even if its specific manifestations are at times grating...

As for some of the recent heat between DTrace and SystemTap recently: I would like to point out that we on Team DTrace were largely silent about SystemTap for nearly three years; it was only what we saw to be dishonesty in the way they presented and used our work that spurned us to comment on their effort. And no, I don't view us calling on others to consider porting DTrace to Linux to be equivalent to us denying SystemTap's right to exist, just as I don't view those who call on Sun to GPLv2 OpenSolaris as denying our right to exist...

I have never said and would never say that it's a "shame" that either SystemTap or Linux exists

Yes, but it is a shame that Linux and OpenSolaris can't share code. License fragmentation is the bane of free software, and it's not in the best interest of the users of either platform. I had hoped that Sun and the Linux community would have had more serious conversations about license compatibility. I had hoped that the parties would have had a mutual appreciation for the value of sharing code.

It's a shame we're not working together. That's all that Andrew really meant.

Funny, the Linux idea of "working together" is to take code under a GPL-compatible license, absorb it, and then not give anything back under the original license. Sounds more like free software cannibalism than "working together."

Yes, but it is a shame that Linux and OpenSolaris can't share code. License fragmentation is the bane of free software, and it's not in the best interest of the users of either platform. I had hoped that Sun and the Linux community would have had more serious conversations about license compatibility. I had hoped that the parties would have had a mutual appreciation for the value of sharing code.

It's a shame we're not working together. That's all that Andrew really meant.

Why should he blame Sun for that, though? They are fully compatible with FreeBSD, OpenBSD, NetBSD, MacOS X, and easily a great many more. Likely you could use their code with just about any non-GPL operating system.

Meanwhile, had they already gone with GPL, 2 or 3, whichever, they would now be compatible with... erm... Linux? How is that better? Just because you could use it in that case?

Why isn't he up in arms to make Linux switch to a license compatible with the rest of the world?

Maybe the difference is not between competition and psychotic monomania as much as between competition and the ability to cooperate. What is the better carrot for development?

OpenSolaris chose competition, and obviously that was a Good Thing(tm) for You. Now that DTrace could benefit from more users all that is available is, still, competition (with SystemTap). I can understand both sides being frustrated with this, and some people are hoping for an obscure middle ground (as in Ahläs post). Others, not believing in the middle ground, are wishing that the original choice of competition before cooperation was reversed (i.e. discontinued OpenSolaris).

I don't think You should take it as something negative, but rather be proud that Your work (and others' like You) trigger such strong emotions. (As long as You didn't have anything to do with the choice of CDDL rather than GPL. That said, I'm still hoping it was a bad quote or at least not meant the way you interpreted it.

As a Linux user, I'd like Team DTrace in "my" boat, but since that's not happening I guess competition will have to be a Good Thing (tm) for SystemTap. Hopefully they will also behave better in the future.

Bryan, as nasty and plain weird as those statements were (again, assuming the story is true), I think that you can interpret them as a compliment. If a kernel guy is saying these kind of things, it means he is scared that Solaris is beating Linux on merits. In Linux Camelot, that is just not supposed to happen.

Regardless, I remain thankful that Sun has opened up such a wonderful OS.