OpenSolaris isn’t a true open-source project, but rather a “facade,” because Sun Microsystems doesn’t share control of it with outsiders, executives from rival IBM say.
IBM Takes Potshots at OpenSolaris
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
… if I am the decision maker at IBM, I’ll sponsor a community to start porting OpenSolaris to IBM’s Power/Cell NOW.
To osnews’ butter: what do you think ?
I know you’re a die-hard supporter of Linux and Free Software in general, and …
I don’t know if you are asking osnews or osnews readers with this “butter” word, but I am personally only a supporter of Greece’s national basketball team and nothing else.
LOL.
Sorry, I am asking for the opinion of osnews reader “butters”:
http://osnews.com/user.php?uid=1405
As an OSNews reader, I personally am a supporter of UConn Huskies college basketball, and nothing else. As a professional system administrator, though, opensolaris is only on my radar as a hobbyist OS, on the other hand I use RHEL and Suse daily at my workplace.
You’re cheering for the wrong Huskies, my friend. UW all the way, baby.
Oh, and ZFS makes me tremble with joy. I’ll probably faint when DragonFly BSD announces their port (due soon, according to Matt).
Rock on OpenSolaris. There’s plenty of room in the OSS pool. Sure, it’s not entirely open yet, but it’s still a hell of a lot of code released, and good code at that. I used to be a Sun hater, but man, their engineers are miracle men (and women).
Butters is good at many jobs.
IBM supports Solaris (not OpenSolaris, though) on System X, which is IBM’s line of x86-based servers. I’m sure you can understand why IBM doesn’t sell SPARC servers, which is the only other architecture currently supported by Solaris.
Porting operating systems to other architectures is nontrivial, especially when the OS wasn’t designed with portability in mind. Judging from the time it took for Sun to port Solaris to x86, it’s probably not especially easy.
Solaris is a very well designed OS. When OpenSolaris was released, I read through the task management and scheduling code, which is generally a good litmus test for kernel design. The Solaris design clearly shows that the design came before the code. There’s a sophistication of design that Linux lacks. It speaks of a handful of very smart people developing an elegant and robust system.
On the other hand, Linux 2.6, especially in the area of the kernel I’m comparing, has some implementation details that are just so simple that it’s brilliant (and fast). The emphasis is on simplicity and efficiency–it strives for clean implementation rather than clean design. It speaks of the combined insight of dozens of developers, where the likelihood of moments of genius is higher.
You know, I don’t really have a problem with the CDDL, and I think it’s the most practical and applicable open source license given the prevailing interpretation of copyright and patent code. One fundamental difference from the GPL is that it distinguishes the original author from subsequent contributors. That’s designed to directly address the question of who’s granting the intellectual property rights to the user. The user receives a license for all covered software, but the license for all but the original software is granted by the distributor. When it comes to patent and other intellectual property provisions, I think the CDDL bests the GPLv3 drafts by a large margin.
The part that gets FSF people in a twist is the file-granular definition of modifications and covered software instead of the linking-based definition in the GPL. This is the fine line between copyleft and viral–whether it is enough that your code will always be free, or do you demand that all code that uses your code be free. The CDDL says “all code in this source file shall be free,” whereas the GPL says “all code in this executable file shall be free.” Draw your own conclusion.
I agree with the point that we don’t need more than one open operating system, but I say, the more the merrier. Software development is as much art as it is science. As more interpretations of this art are shared with the world, we gain new insight that leads to new research and new innovations. Try as they might, they can’t take away our right to combine our insight and ingenuity to create progress.
I’m going to have to give the thumbs down to IBMs comments on OpenSolaris. By any account, both the OpenSolaris and Linux development communities are producing great systems that will continue to thrive in the enterprise. The Linux community has its values and Sun has their’s too. The market will judge them based on their products, not on the openness of their development model. If the OpenSolaris development community is (seemingly) more exclusive, then Sun will bear the burden, not their users.
It’s usually smarter to ignore your competition than to publically attack them anyway…
Edited 2006-08-17 06:06
Porting operating systems to other architectures is nontrivial, especially when the OS wasn’t designed with portability in mind. Judging from the time it took for Sun to port Solaris to x86, it’s probably not especially easy.
Actually it did not take the very long at all to port Solaris from Sparc to x86. Solaris on x86 has been around since atleast Solaris 2.4. There was eithen was a previous port to powerpc (which the community is now revisting). Any percieved lag probably showed at the time, Sun put only a small amount of resources into platforms other than Sparc. Today this is not the case.
There is very little of the Solaris kernel which is platform specific. The fact that the code is very well written and though out makes porting to other platform relatively easy.
Well, x86 support for Solaris seemed to pass in and out of favor amongst the Sun executives a couple times, then it was dropped, and then it was revived. For Sun, the hardest part of the x86 port was deciding if they wanted to support it.
This reflects on the larger question of how “open” the OpenSolaris community really is. If the community proposed an effort to port OpenSolaris to PPC, would Sun be “open” to the idea? Such an effort might take more than 119 contributions, though…
This reflects on the larger question of how “open” the OpenSolaris community really is. If the community proposed an effort to port OpenSolaris to PPC, would Sun be “open” to the idea? Such an effort might take more than 119 contributions, though…
PPC port is already happening. Sun is very open to the idea and quite happy that is proceeding. I think your question is answered.
PPC port is already happening. Sun is very open to the idea and quite happy that is proceeding. I think your question is answered.
How keen they actually are is not particularly clear, the problem is with the nature of OpenSolaris you are entirely dependant on Sun in porting to anything and getting it work.
Just look at OpenDarwin as an example.
Just look at OpenDarwin as an example.
OpenDarwin was in no way affiliated with Apple, in the same way the Fink project isn’t related to Apple. You’re confusing Darwin with OpenDarwin.
On the other hand, Open Solaris has the blessings of Sun and co. This would make it very similar to MacForge, which is an open source project hosted by Apple.
How keen they actually are is not particularly clear, the problem is with the nature of OpenSolaris you are entirely dependant on Sun in porting to anything and getting it work.
No, actually, we are not. I know of nothing that would stop someone from porting to other architectures if they wanted to. Can you name any specific item that would prevent someone from porting OpenSolaris to another architecture?
No, the part that Sun management didn’t get about Solaris x86 was how much it was used and by whom. Once officals from Sun met with the “Secret Six” their entire attitude about Solaris x86 changed and what you see today is the results of that meeting.
I was using Solaris 7 and 8 x86 at that time on machines I had at home and SPARC at work. I put my two cents in when Bill Bradford (of SunHelp) asked for input from Solaris x86 users I provided it, along with a host of others.
For more information on this just Google “solaris x86 secret six”.
You know, I think I actually may have a problem with the CDDL/OSBL.
😉
I guess I’m not sure what (protections and benefits) I get from using Sun’s version(s) of an OpenSource license vs the GPL (or BSD):
http://www.groklaw.net/article.php?story=2005061422100471
This is a relatively old article (with links to even older articles), are things still like this?
Whatever your feeling for Groklaw the issues raised are worth thinking about..
I also have to agree that IBM statement about one proprietary and one opensource OS seems rather ridiculous if you really are committed to open-source.
Time will tell what licenses and OS’s will prevail however. It’s all a good thing..
E.
“… if I am the decision maker at IBM, I’ll sponsor a community to start porting OpenSolaris to IBM’s Power/Cell NOW.”
That’s already happening, regardless of IBM’s wishes. There are projects to port OpenSolaris to PowerPC – http://www.blastware.org/ – and z/Series (mainframe) – http://www.sinenomine.net/vm/opensolaris-zseries
Rather funny considering IBM’s comment that:
One open-source operating system is plenty, though, so there would be no point to making AIX open-source, IBM’s Handy said. “There’s room for a proprietary one and an open one. Once one is open, you don’t need any more,” he said.
So, another words, do what we say, not what we do.
Anyway, what they say isn’t true. The community and governance board together is in control of the OpenSolaris project, even the current temporary one has members that are not part of SUN. So saying that SUN doesn’t share control of it with outsiders is an outright lie.
Sun could do “simple things” to build a real OpenSolaris community if it were serious about doing so, Frye said. “They would push their design discussions out into the forums, so people can see what’s going on,” he suggested.
Actually, they are having design discussions in public. In fact, you will often see internal SUN people responding to the public mailing lists about very large design decisions. Are all of them happening there yet? No. But I sure don’t see public design discussions regarding AIX outside of IBM on a forum.
SUN is still in the process of transitioning their internal processes to external ones in many cases. Anyone with a web browser can see this just by looking at the ARC discussion or program team forums as examples:
ARC forum / mailing list:
http://www.opensolaris.org/jive/forum.jspa?forumID=10
Program Team:
http://www.opensolaris.org/jive/forum.jspa?forumID=77
CAB (Community Advisory Board):
http://www.opensolaris.org/jive/forum.jspa?forumID=17
IBM as usual is full of hot air, they only help when it’s convenient for them.
It’s funny because everyone criticizes SUN for not “getting” open soruce and for not doing enough, even though they have released millions of more lines of code than IBM ever has under an Open Source license. SUN has even been taking many of their “crown jewels” and making them open source. Yet, we see comments from IBM about it not being necessary for AIX, yet I see little criticism of them over that.
As far as SUN not doing enough to build a community? Hah. What community has IBM built? IBM decided to contribute to the Linux community primarily. Whereas SUN not only has done a great deal of work to build not only their own communities, but also to contribute to many other communities as well! It’s poor form for IBM to criticise another for what it does not even do itself.
Nothing like a double standard.
Edited 2006-08-17 04:08
What community has IBM built?
Did you know that IBM wrote and released the Eclipse java IDE in an effort to “eclipse” Sun’s netbeans with an open source IDE? I guess you didn’t and Eclipse is pretty much the de-facto industry standard java IDE now.
Also, I’m not going to count the fact that Linux is “enterprise ready” due to large contributions from IBM and Silicon Graphics. The scalability improvements in the 2.6 kernel are largely due to these two companies. IBM has contributed WAY more to open source than Sun has or likely ever will.
http://news.com.com/2100-1001-249750.html
Granted, Sun has spent millions on usability research for the gnome desktop and they wrote the initial version of it’s famous Human Interface Guidelines, but they still don’t match big blue.
Sun sees Linux as a competitor (which it is) and they created the CDDL so that Linux can not benefit from Sun “open sourcing” it’s code. Also, IBM has donated more patents to the Patent Commons project than any other single company. Don’t believe me?
http://www.patentcommons.org/commons/patentsearch.php?formType=resu…
I do agree with you on AIX, but Sun isn’t near as “open source” as it’s propoganda machine wants you to believe. Sun took text book marketing from Microsoft and turned it on other companies like HP, IBM, and Dell. Remember the ridiculous statement made recently by a Sun exec about how HP-UX should be open sourced and integrated with Solaris because it wasn’t any good? HP is *STILL* leading Sun in Unix server sales, but they keep quiet about details like that.
Also, I’m not going to count the fact that Linux is “enterprise ready” due to large contributions from IBM and Silicon Graphics. The scalability improvements in the 2.6 kernel are largely due to these two companies. IBM has contributed WAY more to open source than Sun has or likely ever will.
Care to back that up with numbers? In terms of sheer code, SUN has contributed more to the Open Source community than IBM. Unless you’re going to argue about who they contributed to, in which case, that’s a subjective qualification.
Sun sees Linux as a competitor (which it is) and they created the CDDL so that Linux can not benefit from Sun “open sourcing” it’s code.
No, they did not. That is pure speculation with NO proof whatsoever. The GPL was one of the licenses considered during the licensing decision process. It was rejected because it did not meet the needs or qualifications of the project.
Granted, Sun has spent millions on usability research for the gnome desktop and they wrote the initial version of it’s famous Human Interface Guidelines, but they still don’t match big blue.
How? It’s nice to say things without proving them. Does IBM have an entire dedicated desktop team that contributes on a regular basis to the GNOME project? What’s that? They don’t? I’m shocked! (/sarcasm)
Also, IBM has donated more patents to the Patent Commons project than any other single company. Don’t believe me?
…and a lot of those patents as were even pointed out by others a few years ago are not even applicable to most software.
Why are they bashing the OpenSolaris community – seems like they’re doing some really good work, getting really good folks involved, too. ZFS and containers rock.
<p>
We’re running Solaris on IBM’s blades, and it runs beautifully. Now I’m wondering if we should revisit that decision… IBM is just feeling lost to me, flailing around sounding like Sun did a few years ago, angry and out of touch.
It’s simple: IBM wishes that Sun (and OpenSolaris) would just go away and leave the server market wide open. In a lot of ways, OpenSolaris is superior to IBM’s commercial AIX offering — and IBM isn’t happy about it; particularly when, as you suggested, customers run OpenSolaris on IBM blades. It would make it that much easier for customers to move over their environments to Sun hardware. IBM is no different than any other vendor: They want to lock you into their offerings. Keeping customers on Linux makes it easier for their marketeers to argue for migration in the other direction. But, as for whether OpenSolaris is really “open”, IBM is splitting hairs. And it’s really counterproductive, IMHO.
We’re running Solaris on IBM’s blades, and it runs beautifully. Now I’m wondering if we should revisit that decision… IBM is just feeling lost to me, flailing around sounding like Sun did a few years ago, angry and out of touch.
I wouldn’t say that, I’d say that IBM is scared; imagine this; IBM currently sells their POWER computers at a loss (along with the future Cell ones) – lets imagine for a moment that Solaris 11 is completely ported to Cell/POWER architecture (including ‘OpenPOWER’) – IBM hope that with making a loss of their initial hardware, they can make it up via the sales of services and middleware.
Now, lets make another assumption that Sun Microsystems come out and say, “hey, this Cell/POWER port is really good, we’re going to officially support it; you can sign up and get the same level of support that we have for the SPARC and x86-64 hardware.
Then you end up with companies, in a mixed environment of Sun and IBM hardware going, “we could replace AIX on our machines, move all out support and maintainance to one company, get all out middleware from Sun and Oracle (assuming they use Oracle DB) and come out better off with lower costings.
IBM will look at this as a customer lost; they’re not going to buy any services because IBM refuses to participate and help port OpenSolaris to the POWER architecture, and IBM will also lose out on middleware sales because, well, according to their last PR announcement, “Solaris isn’t ready for prime time” – so effectively their whole ‘loss of hardware, profit on middleware and services model’ will fall over.
I for one see OpenSolaris a step in the RIGHT direction of which IBM should take advantage of; an opensource, fully UNIX certified operating system, without any of the GPL related issues that go with it; there is nothing stopping IBM from creating their own workstation/desktop/server distribution using OpenSolaris, offering services and middleware.
The simple fact is, IBM is suffering from the same NIH syndrome which Sun suffered from 10 years ago, with the SPARC bigots running the show, and REFUSING to acknowledge that the SPARC architecture was getting its ass handed to it on a platter – 12 way SPARC machines getting beaten by quad Xeon machines running Linux/Windows.
ps. What GPL issues? binary drivers for one thing; yes, I can’t stand them, but at the same time its a necesssary evil in life; I’d love for everything to be opensource and licenced under the most permissive licence under the sun, but we all know that we have people in higher places concocting all manner of stories as to why they can’t opensource the said driver – reminds me of the old “Yes Minister” episodes with the steps used to stall a minister from making a decision; firstly claim that its impossible, then claim that there has been a study and say its not feasible, then if they still persist, claim that there are legal, possibly ‘national security issues’ at stake, then when all else fails – which normally doesn’t happen, but if so, make up another excuse like an ‘independent study’.
Edited 2006-08-17 06:38
Marvellous analysis, kaiwai. Brilliant conclusions, of which I particularly liked: IBM will look at this as a customer lost; they’re not going to buy any services because IBM refuses to participate and help port OpenSolaris to the POWER architecture, and IBM will also lose out on middleware sales because, well, according to their last PR announcement, “Solaris isn’t ready for prime time” – so effectively their whole ‘loss of hardware, profit on middleware and services model’ will fall over.
If Sun is smart, they’re going to push OpenSolaris even more and further. The quality of OpenSolaris is such that it will penetrate ever more business ecosystems and mindshare is growing.
If Sun is smart, they’re going to push OpenSolaris even more and further. The quality of OpenSolaris is such that it will penetrate ever more business ecosystems and mindshare is growing.
And what to know the really cool part? even with the various other distributions out there, like GNU Solaris and Sun Solaris, the OpenSolaris communities all benefit from the input that each other put into producing a distribution.
It has all the benefits of FreeBSD, but with corporate backing in the way of resources, R&D and so forth.
Having had a look at GNUSolaris (currently at Alpha 5), all I can say is that I am incredibly excited about the eventual release; a Solaris distribution that will be 100% on par with the desktop oriented linux distributions out there – as a side issue, yes KDE and GNOME will both be supported on it.
I think you will find IBM financial commitment to OSS at least 5-10 times the value of Suns. Remember IBM makes money on services. They even offer Solaris support from their staff. If comparison to Linux Open Solaris is more proprietary.
Begin off topic IBM rant.
IBM has considered open sourcing large parts of AIX. The problem has always been that large parts of the VM and file systems are owned by a sub contractor. The Linux JFS is a reimplementation.
Sun is barely alive. The AMD switch may have saved them but they are like Apple they are a hardware company. Software is the value add to their overpriced hardware.
IBM is the same but the make all the parts, integrate, and deliver. Current sparc hardware cannot touch power series systems. On the low end the low power smp is a market ibm is not in at this point.
People who use AIX and Solaris are not in the market for Linux/open Solaris. Their requirements are totally different. It isn’t how much money can I save on this machine. It is over the five year life time what are the real costs and real risk for this DB which is worth 5 million dollars a month to us.
Open Solaris is just a branch all the Sun people who know they have lost can hang on. Hell HPUX on Itanium is a more powerful platform.
Generally speaking IBM _was_ trying to move AIX more main stream. But then the unexpected happened. Itanium failed (IBM kill the port of AIX). Sun fell behind the performance curve. So at 8 CPUs and greater there are no competitors with believable life cycles left.
Only Intel which keeps trying to creep up. But the P series platform has what 14 years on them. So if you are not a computer company and want to buy from a major supplier what are you going to buy?
And all of what you say is based on …? Is that why IBM can’t make up its mind what to support, I have read two articles in the last 8 months about IBM “recommiting” to AIX (search here and you can find them both) and their recent announcement about throwing more money at Linux! So what are they trying to say, from where I sit it sounds like they have a problem. When I talk to a group of SA’s I don’t hear anything about AIX (and yes, I have administered AIX).
And HP-UX, talk about a company in search of a clue! HP basically threw their enterprise strategy out the window and bet the farm on Itanium! As a result they have alienated a bunch of their customers who see an expensive migration from PA-RISC to either what HP has to offer (x86 or Itanium) and you eat the costs associated with migrating apps, or go somewhere else.
I sat in on a meeting with several HP representatives and if what they said in that meeting about spending more money on Linux than HP-UX was an indication of the future of HP-UX, then I would start looking at other options. Also, we buy a lot of HP (Compaq) gear, right along side the Sun hardware.
I think that pSeries hardware is grossly overpriced for what it provides compared to Sun stuff, and it also happens to be some of the quirkiest machines around (just love crazy 8’s). And lets not mention those overpriced services (yes I know about IBM Global Services), no wonder they have “money to burn”. If the last project I worked on was any indication of the quality of IBM Global Services, I would want my money back!
And the “Sun is dying” nonsense, do us a favor and can that. That statement is about as silly assed as “BSD is dying” trolls on Slashdot.
I think you will find IBM financial commitment to OSS at least 5-10 times the value of Suns. Remember IBM makes money on services. They even offer Solaris support from their staff. If comparison to Linux Open Solaris is more proprietary.
Begin off topic IBM rant.
IBM has considered open sourcing large parts of AIX. The problem has always been that large parts of the VM and file systems are owned by a sub contractor. The Linux JFS is a reimplementation.
Sun is barely alive. The AMD switch may have saved them but they are like Apple they are a hardware company. Software is the value add to their overpriced hardware.
IBM is the same but the make all the parts, integrate, and deliver. Current sparc hardware cannot touch power series systems. On the low end the low power smp is a market ibm is not in at this point.
People who use AIX and Solaris are not in the market for Linux/open Solaris. Their requirements are totally different. It isn’t how much money can I save on this machine. It is over the five year life time what are the real costs and real risk for this DB which is worth 5 million dollars a month to us.
Open Solaris is just a branch all the Sun people who know they have lost can hang on. Hell HPUX on Itanium is a more powerful platform.
Generally speaking IBM _was_ trying to move AIX more main stream. But then the unexpected happened. Itanium failed (IBM kill the port of AIX). Sun fell behind the performance curve. So at 8 CPUs and greater there are no competitors with believable life cycles left.
Only Intel which keeps trying to creep up. But the P series platform has what 14 years on them. So if you are not a computer company and want to buy from a major supplier what are you going to buy?
Your rant is full of shXt. I just love your over priced hardware statement. You are just simply out of touch. Maybe you should look at the price lists again. You might get a shock.
I think you will find IBM financial commitment to OSS at least 5-10 times the value of Suns. Remember IBM makes money on services. They even offer Solaris support from their staff. If comparison to Linux Open Solaris is more proprietary.
Like what? Sun was the first to provide (thanks to the blackdown port) and maintain a copy of Java for Linux; they spent $52million to buy out StarOffice, spent unknown sums coming up with the OpenDocument format and seperation of the applications, which we’re all take for granted when running a UNIX like variants out there.
Then there are the contributions to GNOME, Netbeans, GTK, Linux by way of NFS support, then there is the big exchalada, the opensoucing of OpenSolaris, with the closed, licence hindered portions being replaced with superior re-written opensource versions. How can you claim that OpenSolaris more ‘proprietary’?
Linux is like a bare back orgy, you don’t know who is going to f*ck you, and you’re stuffed if you catch something – the viral nature of GPL; Atleast with OpenSolaris, you can go along to this swingers establishment, contribute, take, and have a damn good time, knowing that you’re not going to ‘accidently’ get comtaminated or have legal issues hanging above your head as with the situation of the Linux GPL and binary drivers.
IBM has considered open sourcing large parts of AIX. The problem has always been that large parts of the VM and file systems are owned by a sub contractor. The Linux JFS is a reimplementation.
The JFS implementation is a port of the JFS that appears in the old OS/2 Warp eCommerce Server.
Sun is barely alive. The AMD switch may have saved them but they are like Apple they are a hardware company. Software is the value add to their overpriced hardware.
Excuse me, how are their Opteron machines more expensive than the competition? I’ve looked on the Sun Catalogue, and everyone of their Opteron based products are either cheaper or atleast on par with the prices which Dell offers their Intel EMT64 machines at – hardly what I’d call ‘over priced’.
Regarding their SPARC machines; again, what rock have you been hiding under? the Niagara processor is going gang busters in terms of sales thanks to its ‘raw throughput’ approach which Sun is concetrating on, and shock bloody horror, Sun STILL outsell IBM in terms of UNIX systems; bigger ecosystem, more customers, and more units shipped.
IBM is the same but the make all the parts, integrate, and deliver. Current sparc hardware cannot touch power series systems. On the low end the low power smp is a market ibm is not in at this point.
Then obviously you have *VERY* limited knowledge to the issues regarding the POWER architecture; cute benchmarks don’t show some of the short comings of it – IIRC Bascule and Mario in the past have already explained these issues.
With all the talk about OpenSolaris, you’d think it really was SOMETHING. However, you go to the OpenSolaris Site – where do I download a CDROM iso image to install the thing? Oh! I’m supposed to first go get a Nevada iso set from Sun’s web site (all the while giving them buttloads of information), install that, then download some mystical code blurbs from the OpenSolaris site, let it compile for a few hours, and HOPE it works, because if it doesn’t, you’re going to have a non-bootable system – Believe me, I’ve tried, and I’m no dummy when it comes to compiling stuff – I’ve been heavily involved with the LinuxFromScratch project for a number of years. But yet, I’ve tried to put OpenSolaris on a spare X86 box, as well as a Sparc E420R and a Ultra30, none of them would actually boot up when the long compile process was complete.
But yet, I’ve tried to put OpenSolaris on a spare X86 box, as well as a Sparc E420R and a Ultra30, none of them would actually boot up when the long compile process was complete.
The process at the moment does have a few hairs, but I replying to you from an Open Solaris compiled install (with ZFS root) that never touched the Sun Installer. It is certainly do’able and bootable.
With all the talk about OpenSolaris, you’d think it really was SOMETHING. However, you go to the OpenSolaris Site – where do I download a CDROM iso image to install the thing?
As SUN says on the Downloads page, “Solaris Express is a binary release for customers. It’s Sun’s official release of the OpenSolaris bits as well as additional technology that has not been released into the OpenSolaris source base.”
However, where to download an OpenSolaris distribution? That’s easy, there’s a list right on the OpenSolaris website:
http://www.opensolaris.org/os/about/distributions/
Not hard to get to either, all I had to do was:
1) goto http://www.opensolaris.org
2) Click downloads
3) Click the ‘Distributions page’ link.
So, where is it you ask? The answer is simple, on the downloads page.
Edited 2006-08-17 05:50
If you’re really interested in doing some
OpenSolaris development, I’d suggest sending
a detail report of what you tried on each of
the three machines you listed to the OpenSolaris
help forum
http://www.opensolaris.org/jive/forum.jspa?forumID=31
Yes, it would be wonderful if you could get everything
you need from the initial installation but at
the moment, the source (plus some binaries which Sun
is not able to distribute source) can be easily
downloaded. The build process, as with most large
source projects, has its own unique steps but
is straightfoward.
The point is that there is a community out there
who would love to help you get started – we only
ask you to participate and keep an open mind.
I agree; on this basis, you’d be able to say that AmigaOS is still a success, even although it runs on a tiny amount of machines.
I’ll believe OpenSolaris is a success when I can run it on my (bog-standard 3 year old) desktop; and I’ll believe the GPL is business-unfriendly when it declines to 10% of the open-source market (unlike the 45-50 it has presently, according to some) and Red Hat go insolvent.
As someone already noted (but you failed to read): did you try to install Solaris Express, the binary distribution of Opensolaris?
As someone already noted (but you failed to read): did you try to install Solaris Express, the binary distribution of Opensolaris?
Can you compile everything released as a part of OpenSolaris and get a working system like Solaris Express at the end of it?
No. As a result, OpenSolaris is not a real open source project. As I’ve pointed out, just look at the difficulties OpenDarwin got into.
I modded you up 1 from 0 because you are correct. From what I understand, Nexenta is still linking everything against Sun’s CDDL c libraries which are GPL incompatible.
OpenSolaris is a very good marketing ploy + an easy way for Sun to look good while still holding onto the crown jewels. It also allows them to offset development costs slightly.
What could Sun do to have a “real” open source project? Put OpenSolaris in a public cvs repository and give outside contributors (read non-Sun employees) full commit acess to it. Make the engineering and design decisions on a public and non-moderated mailinglist while being open to outside ideas. Do you see this happening? No you don’t.
Sorry Sun, no cookie for you.
Edit: Fixed a typo
Edited 2006-08-17 13:36
What could Sun do to have a “real” open source project? Put OpenSolaris in a public cvs repository and give outside contributors (read non-Sun employees) full commit acess to it. Make the engineering and design decisions on a public and non-moderated mailinglist while being open to outside ideas. Do you see this happening? No you don’t.
Yes you do. Design discussions are being held on the public mailing lists. Public repositories with outside commit access have been part of the plan all along, and the infrastructure for them is being actively worked on and tested. See http://www.opensolaris.org/os/about/roadmap/ for the roadmap, and look at the SCM items (the schedule is subject to change, as are all schedules, but the items are in there and being worked on)
Umm, sorry you are wrong, it is happening. If you don’t see it, it just means you are looking in the wrong place, not that it isn’t happening – try http://www.opensolaris.org/os/community/tools/scm/ As one of the engineers working on OpenSolaris, I can assure you it is most definitely happening – I’m working on the SCM web console as I write this.
I modded you up 1 from 0 because you are correct. From what I understand, Nexenta is still linking everything against Sun’s CDDL c libraries which are GPL incompatible.
Nexenta has actually worked out everything with the Debian and FSF and they are in total compliance. So this is wrong.
OpenSolaris is a very good marketing ploy + an easy way for Sun to look good while still holding onto the crown jewels. It also allows them to offset development costs slightly.
On what basis can you say this? This is pure speculation.
What could Sun do to have a “real” open source project? Put OpenSolaris in a public cvs repository and give outside contributors (read non-Sun employees) full commit acess to it.
As opposed to a “fake” open source project? Really, this is getting old. As far as commit access for outside contributors, this was announced as part of the roadmap and plan a year ago. It’s still on the agenda, and it’s being worked on right now. It’s still on schedule for happening too. It is really annoying when people run around talking about things they don’t know about.
Make the engineering and design decisions on a public and non-moderated mailinglist while being open to outside ideas. Do you see this happening? No you don’t.
You haven’t been looking very close then. This is happening right now. Go read the program-team, CAB, ZFS or other forums. You will see a lot of internal SUN employees talking about and discussing design and other decisions openly in public. The lists are only moderated to prevent spam, all legitimate messages are let through. Even the linux-kernel mailing list is moderated to a certain extent!
The lists aren’t even really moderated– you just have to sign up and be an actual mailmanified member of the list to post… that’s to prevent the spambots.
Trust me, I’m the “admin” for one of the lists, and we get about 20 spams a day, and it’s really annoying.
Can you compile everything released as a part of OpenSolaris and get a working system like Solaris Express at the end of it?
Yes, you can compile all of the source that is released for OpenSolaris and get a working system. How do you think SchilliX happened?
No. As a result, OpenSolaris is not a real open source project. As I’ve pointed out, just look at the difficulties OpenDarwin got into.
Oh, but it is a *real* open source project with a *real* open source licensed approved by the *real* Open Source Initiative organisation.
Or are all of the “working” OpenSolaris distributions “fake,” like GNU/Solaris http://www.gnusolaris.org/gswiki or SchilliX, etc?
Was SuSE fake back when YaST was under a different license? The list goes on.
Are Linux systems that use binary blobs for their WiFi drivers or Video drivers “fake”?
If not, then I don’t see how OpenSolaris is.
No, and a binary-only OS, or a supposedly “open” OS with binary-only bits in, is not “open”. It is at best “ajar”.
I guess Linux distributions with nVidia / ATi / Intel binary blob drivers aren’t “open” either then…
There is a BIG difference between having to include binary blobs because that’s the only way to provide functionality in other companies’ products, and making sure that only people who take your own binary blobs have the best OS experience.
But no, distros which include proprietary whatever are NOT fully open, which is exactly why most distros do not include proprietary modules/software by default.
There is a BIG difference between having to include binary blobs because that’s the only way to provide functionality in other companies’ products, and making sure that only people who take your own binary blobs have the best OS experience.
So which are they then? The only way I can get 3D functionality on my hardware with some of those Linux distributions or network functionality is with those blobs. So who is doing what again?
That’s because the hardware vendors won’t release the specs. Sun’s piecemeal attitude to open source, by contrast, is because, like Microsoft, it doesn’t want OSS around; but they realise that since their technology, being a Unix, is more vulnerable to the Linux/BSD onslaught, they want commodity hardware and FOSS to always play second fiddle to (their) proprietary hardware and software.
That’s because the hardware vendors won’t release the specs. Sun’s piecemeal attitude to open source, by contrast, is because, like Microsoft, it doesn’t want OSS around
That is also a lie and pure speculation on your part.
The only things that are not open source in OpenSolaris are because SUN does not have the rights to open them, or because they are still investigating / securing the rights to do so. It’s hardly piecemeal. Have you actually compared the amount of open code in OpenSolaris to closed binaries? The majority is overwhelmingly open. Very few things cannot be opened.
Edited 2006-08-18 01:30
I’ll believe OpenSolaris is a success when I can run it on my (bog-standard 3 year old) desktop
Your measure of success might differ from the Sun’s.
I’ll believe OpenSolaris is a success when I can run it on my (bog-standard 3 year old) desktop
Have you tried installing it on your bog-standard 3 year old desktop? The recent Nevada builds (recent being within the last about a year or so) will install on pretty much anything fairly well. The worse problem so far is networking, you generaly need to get a driver off the net, but a simple google search finds them.
Or, you can try one of SchiliX, BeliniX, or Nexenta. All three include many of the extra drivers that Sun doesn’t (potentially, can’t for the moment) and work very well on bog-standard hardware.
Yes, I did try it on my bog standard 3 year old desktop; why else would I talk about “when I can install it” on such as if it were a (possible) future event? I’m downloading Solaris Express at the moment to see if I have any more success with it than with the other three you mentioned.
Drivers aren’t much good when the thing won’t even boot. Ironically, it had more success on my new laptop than that old box.
I’m waiting for IBM to open source OS/2.
…the day hell freezes over and Microsoft’s bankrupt…
Where’s OpenAIX or OpenOS/2? Maybe IBM should write a new VM, language, and libraries so that Eclipse has a new platform to ride on top of, since apparently Eclipse is what is important and somehow we should just forget about Java.
This is nothing new. IBM doesn’t care about open source and either does Sun (except if it can help the bottom line). But at least Sun is putting their money where there mouth is – even if their motivations don’t coincide with what they say in public.
IBM top executives should reign in these smack-talking cowboys. Sun should too when Gosling and others talk smack about SWT.
What about open source DOS/VSE, MVS and VM, too?! 😉
VM at one point WAS open source – one can legally obtain old copies of VM to run in Hercules, the S/3×0 emulator.
OS/2 apparently has many IP encumberances, many of them involving Microsoft. While some of OS/2’s tech has been released as open source (JFS2 comes to mind), the chances of the rest of OS/2 being released are low due to those
issues.
I don’t know if AIX has similar issues or not…?
IBM would have the same (if not more so) legal issues as Sun did in creating OpenSolaris. Of course the Linux fans seem to forget that in their zeal to point out the shortcomings of OpenSolaris.
Try “in their zeal to point out that Sun is being economical with the truth” about open-source. Sun’s relationship with FOSS is and always has been schizophrenic at best (witness NeWS vs. NFS).
Actions speak louder than words, as long as AIX is closed they kind of lose their moral authority on the issue don’t you think?
This is not directly related to OpenOffice.org, but at least it is related to sun.
While (in my opinion), OpenSolaris is doing great in getting new contributors, OpenOffice is a completely different matter.
I had to develop some plug-in for OpenOffice for my company. First
1) The developer book is really to big – and not well explained. Try to search in it how to handle events from OpenOffice! (and no, addEventHandler() is generally not enough!)
2) Asking question doesn’t help. No one takes cares about you on the mailing list. That’s plain dumb, because I am a potential contributor: I’m already working on the code, so it would be only a little step more to contribute.
3) The committers are almost only from sun. No other committers (who wonders! See point 1) and 2)!)
OpenOffice is already complex to understand – but that would be all right if at least there would be better documentation and a more helpful mailinglist.
Altogether the OpenOffice engineers made my task very difficult. Handling events from Openoffice.org took me 2 weeks. (I will publish how to do it on the code-snippet wiki, at least other people wont lose so much time as I did).
The development of OpenOffice could be faster if the community were more open. Look at eclipse, or at Jena, the ontology framework! You always get help from the mailing list. Documentation is great. IRC is a great help to. (Why is there no #OpenOffice-api on freenode?)
We certainly will use an other editor instead, possibly by using eclipse/GEF. And that is sad, because OpenOffice.org matched our need better .
I have no experience with programming OO.org, but OSNews and other sites have run stories on the problems with the codebase of OO.org. Basically, it’s full of cruft from nearly a couple decades back. I think at one time, in the not so distant past there was some even some late 80s assember code in there.
So just having a “more open community” doesn’t do much good if the codebase is so dense and crufty that you basically have to become a fulltime OO.org engineer to do anything significant.
But of course this is the problem with many open source projects. You’ve got thousands and thousands of lines of code, no documentation on the code, which result in high barriers to entry – especially when many people want to contribute in their spare time. You have to hand it to the eclipse community that has produced a number of books on Eclipse plugin writing.
IMHO, what we need is to break out of the C/C++ ghetto, move onto higher level languages, and get much, much better tools that help us understand code. Easier said than done of course.
In fact, OpenOffice has a lot of binding for other, more higher level language, thanks to UNO. It is possible to write OpenOffice component into Java and Python. E.g., The database component is mostly written in Java. So the developer must not use C++. There also exist some bindings which allows access, but not writing components for StarOffice Basic, OLE Automation and the even .NET Common Languare Infrastructure
I don’t really think that UNO is bad. It tries to accomplish a lot, and it is quite good in it. The problem is the community and the documentation.
The Java bindings are bad. They are very low level. It is always OK to have low-level libraries, but there should also be a more higher-level wrapper around them. These libraries are contained in jar files which for most parts have no available javadoc. That’s a plain then programming with tools like eclipse. Instead you have always to look at the idl documentation (IDL is an interface definition language, similar to corbas idl). See at http://api.openoffice.org/docs/common/ref/com/sun/star/module-ix.ht…
That makes programming with code completion much harder, since I have to guess by the name of the method what the method is supposed to to. And that is in the case of OpenOffice.org not evident at all.
And, as I said previously, the only available documentation is bad. The developer guide is too long, and has bugs. Some example even don’t compile. It is not possible to change the developer guide and fix bugs, since it is the developer guide is not open source.
As for OpenSolaris: I didn’t look into the OpenSolaris community, so I can not judge there. OpenOffice was for me the proof that it is possible to have a non open (community!) around an open project. Lets hope OpenSolaris does that better.
p.s.: English is not my first language, sorry for the mistakes.
Really guys, if Sun was really interested in making this OS open source, they should have first developed a good installer. Installing solaris as of now is a pain in the a**.
Maybe the real reason for the word open in opensolaris is just to get some good points from the open source community and nothing else.
Do you know, the solaris man pages are *not* open source. That is why distributions such as nexenta do not ship with the man pages.
The manpages are being actively worked on at present – targeted for 09/2006. See http://www.opensolaris.org/os/about/roadmap/
Not to defend IBM but I think that they are stating that an OS doesn’t have to be open source to be good, eg. AIX, but if it is going to be marketed as open source it should be not only open but a community project in order to take advantage of open source. So in this light they are not being hypocrites as the open source OS (Linux) that they support is fully a community project and they are just a community member, while the commercial UNIX they support (AIX) is totally closed and corprate and there is nothing wrong with that either.
Personally I am not sure as to what good this perspective does for IBM.
In years past IBM performed all of the necessary R&D and drafted numerous patent applications for “A method to inspire fear, uncertainty, and doubt into customers about the viability of selecting solutions made by competitors of IBM.” They would have filed it too, if it wasn’t for those meddling kids.
Whether IBM is right or not that OpenSolaris has a development community, OpenSolaris is true Free Software.
Free Software is not about a development method but about a way of licensing software. Free Software can build in a community process and in a in-house process as proprietary software can be developed in a community or in-house. It’s not the development method which makes something Free Software it’s the license.
Sad to see that even such a big company with such a big “linux-centre” like IBM doesn’t really understand Free Software.
From what I understand, Nexenta is still linking everything against Sun’s CDDL c libraries which are GPL incompatible.
As far as I know, that’s the case as well. The only reason why it hasn’t blown up is that Sun hasn’t complained.
From what I understand, Nexenta is still linking everything against Sun’s CDDL c libraries which are GPL incompatible.
As far as I know, that’s the case as well. The only reason why it hasn’t blown up is that Sun hasn’t complained.
Sun didn’t and others seem don’t matter.
http://lwn.net/Articles/159248/
http://lwn.net/Articles/159267/
you can simlpy google for oher complaints
Was about the interpretation of a half-sentence in a clause in the GPL, which the FSF has consequently clarified, such that GPLv3 will explicitely allow such linkage, and the FSF has made clear that they did not intend or interpret the wording of the GPLv2 to prohibit that.
It’s in the Nexenta FAQ, now, too.
That being said, the general incompatibility of the CDDL with the GPL keeps a lot of people from working with CDDL-licensed code. I have no idea if Simon Phipps is working on that in the context of GPLv3 drafts, but someone from Sun definitely should do something about it.
I realize people like to criticize the GPL as being viral, but this overlooks the fact the nature of the license was the single biggest enabler for encouraging third-part commercial development.
If IBM chose to support and distribute code improvements to OpenSolaris, the CDDL would require them to assign copyright to Sun and give Sun the freedom to ultimately do whatever they wanted down the road, including revising the license and closing off the code.
This is where the definition of free gets hazy. To a user, there’s little difference. To a contributor, particularly a for-profit one, there is a world of difference between anything BSD/CDDL related and GPL related. The quid pro quo nature of GPL is the single reason otherwise competitive companies collectively contributed billions of dollars worth of resources, development, code and patent indemnity. This would never happen under a less restrictive license.
So to IBM’s point, despite the blustering, OpenSolaris isn’t really an open project in the sense of linux and doesn’t encourage outside development beyond a community of outside developers that may not be as concerned with the licensing any more than they would be with a BSD project. So in a sense, opening Solaris accomplishes little overall other than some interesting porting or implementation projects, because the pace of Solaris development will always be tied to Sun’s investment in it, there’s no way there will be substantative third-party development for Sun to leverage, so really, what’s the point? It is smoke and mirrors.
Having said all that, don’t get me wrong, I applaud Sun for taking the step, even if it doesn’t provide some sort of paradigm shift in the industry, open code is never a bad thing. I just don’t think it’s as big a deal as it’s being made out to be.
I also remember there was some issue with the patent provisions in CDDL that potentially left users at risk, considering the cross-licensing between Sun and MS, but IANAL so can’t really comment on that other than to say people much smarter than me had opinions and concerns.
Just my 2c…
If IBM chose to support and distribute code improvements to OpenSolaris, the CDDL would require them to assign copyright to Sun and give Sun the freedom to ultimately do whatever they wanted down the road, including revising the license and closing off the code.
I’m sorry, I just can’t let that statement go unchallenged – that’s complete nonsense. If Sun could claw back the code under the CDDL, the license would never have been approved by OSI as an Open Source license – http://www.opensource.org/licenses/cddl1.php To reinforce the point, here’s what the CDDL FAQ (http://www.opensolaris.org/os/about/faq/licensing_faq/#no-source) says:
Can Sun ever take away the OpenSolaris source code?
No. The code is available to the community forever.
Disclaimer: I work for Sun on OpenSolaris.
Edited 2006-08-17 15:40
I’m sorry, I just can’t let that statement go unchallenged – that’s complete nonsense. If Sun could claw back the code under the CDDL, the license would never have been approved by OSI as an Open Source license
No I’m sorry, that is complete and absolute nonsense and you’ve either misunderstood, or deliberately misunderstood. Don’t use the cop-out of OSI approval either. Every man and his dog’s license seems to have been approved by the OSI, which is why the OSI has been looking to cut down dramatically on the number of licenses.
When you contribute code to OpenSolaris, licensed under the CDDL, copyright is assigned to Sun. Although the code may remain open, it does not stop Sun from taking your code, relicensing it in any way it likes, and using it in closed source and proprietary software. As such, it doesn’t compel them to keep contributing either. A contributor can also do what they like with their contributions, but the huge problem is that Sun is the copyright holder of your code and can relicense it whenever it pleases. It gets even murkier when the issue of things like patents arises with Sun’s removal of the sections of the MPL which deals with disclosure of patents.
Then there’s the issue of sharing, which is what the open source community is all about. Being able to lean on others, and share alike, and cut costs and development time. There is something called GNUSolaris where they look at incorporating GNU and GPL software into Solaris – jut not the other way around.
Can Sun ever take away the OpenSolaris source code?
No. The code is available to the community forever.
Given that Sun can relicense the OpenSolaris code, including all contributions, the practical answer to that question is yes.
> Every man and his dog’s license seems to have been approved by the OSI
I believe it is customary to say “everyone and and his dog” unless you specifically mean that women and their dogs are not getting their licenses approved.
I believe it is customary to say “everyone and and his dog” unless you specifically mean that women and their dogs are not getting their licenses approved.
The phrase is ‘every man and his dog’ – like it or not ;-).
Edited 2006-08-17 19:20
Indeed. “Everyone and his dog” would make no sense since “everyone” is not gender specific and “his” is.
Every man and his dog’s license seems to have been approved by the OSI
I believe it is customary to say “everyone and and his dog” unless you specifically mean that women and their dogs are not getting their licenses approved.
You have very strange customs. I have never heard that being said by anybody. How do you re-phrase “Man’s best friend” – refering to a Dog (both male and female)?
The word ‘man’ actually has two meanings. One is specifing the male species of human. The other is a general term for both male and female.
> When you contribute code to OpenSolaris, licensed under the CDDL, copyright is assigned to Sun.
Is this true? I always assumed it became joint copyright.
No, the it’s not true. The technical term is “Bullshit”.
The CDDL is DISTINCT AND SEPERATE from the OpenSolaris Community project. No where in the CDDL license document is there any mention of sharing or granting copyright to Sun or anyone else.
There is NOTHING that stops ANYONE from taking the CDDL licensed source code available on the OpenSolaris web site, and making any changes whatsoever that they like, and publishing those changes whereever they wish.
Nothing. Zilch. Nada. They CAN NOT stop you.
In fact, the CDDL REQUIRES you to publish any changes to files that are under the CDDL.
HOWEVER, just like Linus can tell you to “piss off” for any change you want to submit back in to the “official” Linux Kernel, the OpenSolaris PROJECT has their own guidelines regarding ACCEPTANCE of code in to THEIR project.
One of those conditions is to SHARE copyright with Sun.
If you wish to work on the OpenSolaris code tree within the OpenSolaris community as managed by Sun, then that is one of the conditions for them accepting code back in to the tree.
If you want to make a change to OpenSolaris, and NOT share copyright, then make your change, publish it whereever you want: publish a patch, publish the files, publish the entire code tree as BobOS. Build your own community around it and sync up with Suns community whenever it pleases you.
The CDDL and the copyright grant are UNRELATED, and a mere detail of the implementation for the community that Sun is putting together. Don’t like it? Grab the code and make your own community.
OpenSolaris is Open Source.
Can Sun ever take away the OpenSolaris source code?
No. The code is available to the community forever.
No. The code is available to the community forever.
Given that Sun can relicense the OpenSolaris code, including all contributions, the practical answer to that question is yes.
Isn’t this what IBM is referring to?
Given that Sun can relicense the OpenSolaris code, including all contributions, the practical answer to that question is yes.
Isn’t this what IBM is referring to?
Yes, because IBM can’t relicense OpenSolaris, but Sun can.
Just like we have Linux distro’s today that are not truly Open Source, because they’re bundling binary drivers, or licensed codecs, or whatever. Solaris (Suns released and supported operating system based up OpenSolaris) is in much the same boat.
There are some “binary bits” within OpenSolaris at the moment that are a legacy of its closed roots that they are trying to get more and more open, but they released the binaries to make the whole of OpenSolaris as a complete system more viable.
Sun wants to share copyright so that they can incorporate the community changes in to their Solaris code base that they support, and Solaris is released under a different license overall than OpenSolaris is because of other proprietary bits that they’re pretty much required to bundle for the machines and hardware that they support.
With shared copyright, Sun may take a file from the OpenSolaris community project, relicense it, make changes to it, and distribute binaries from it without having to reflect those changes back to the community — just like the original author can, since they share copyright.
It’s simply a business reality for Sun to keep Solaris on the cutting edge as an OS and for their hardware. Not all vendors are in to letting their IP out in to the OSS world, so if Sun needs to make a change to a file to support a piece of hardware, they’re going to retain the rights to do that and that’s why they insist on a shared copyright to give them that flexibility.
But there is nothing stopping IBM or anyone else from doing whatever they wish with the source code and granting nothing to Sun over what the CDDL already grants. Just don’t expect Sun to incorporate those changes in to the OpenSolaris version of the codebase, since they use that to build their internal Solaris releases upon.
With due respect you are conflating two issues here – just because there are lots of OSI-approved licenses it doesn’t follow that any of them isn’t open. Anything approved by OSI must adhere to their definition of
Open Source (http://www.opensource.org/docs/definition.html), and as the CDDL is certified by OSI it therefore qualifies as an open source license, at least by commonly-accepted standards. As you point out, the CDDL is derived from the MPL, which is accepted as being an Open Source license. The changes made are explained in http://www.sun.com/cddl/CDDL_why_details.html, and that has an explicit discussion of why the patent clauses were modified.
To come to your second point, it’s actually a joint copyright assignment – a practice that is also shared with Apache and the FSF. There are already several good discussions of why things were done this way and exactly what it entails, rather than repeat them all here are some relevant links:
http://www.opensolaris.org/os/about/sun_contributor_agreement/
http://www.netbeans.org/about/legal/ca.html
http://weblogs.java.net/blog/driscoll/archive/2005/07/newest_concer…
I’d also like to address what seems to be your meta-point (apologies if I am misinterpreting) – that the CDDL is some sort of Sword of Damocles that Sun is forcing contributors to hold over their heads. If that really was the case, can you think of any quicker way of killing the OpenSolaris community? What would Sun gain from such a course of action?
With due respect you are conflating two issues here – just because there are lots of OSI-approved licenses it doesn’t follow that any of them isn’t open.
The terms as to what constitutes an OSI compliant license are so vague there have consequently been many licenses that have been OSI approved. This has now got to such a bad extent that the OSI is certainly considering chopping the number of licenses, because no one has any idea what ‘open source’ is. Nobody has any real clue what is common between each license, and what the pitfalls are.
The changes made are explained in http://www.sun.com/cddl/CDDL_why_details.html, and that has an explicit discussion of why the patent clauses were modified.
Which reads:
“The required notices in the MPL regarding third party claims and patents (formerly in MPL Section 3.4(a) and 3.4(b)) have been eliminated; they seemed overly burdensome and likely to prevent wider acceptance of the license by by the community. Additionally, none of the other major open source licenses (e.g., GPL, BSD, CPL, OSL) require such disclosures.”
Which is very, very flimsy indeed. The FSF or some other non-profit organisation is not engaged in a patent license agreement with people like Microsoft. What it does eliminate is the need for Sun to notify other contributors if someone else has a claim over the code used. It’s simply a matter of trust really, and not something anyone can go for unless the cards are on the table.
To come to your second point, it’s actually a joint copyright assignment – a practice that is also shared with Apache and the FSF.
Many open source projects are typically not the owners of copyrights of contributions to them, such as the Linux kernel. In the case of Apache and FSF projects, they are two non-profit organisations with many vested interests in them. Sun is a for-profit company with a large amount of proprietary and closed software you would effectively be contributing to, for nothing.
Keep in mind also that Sun holds copyright (albeit jointly) over all of the OpenSolaris codebase as a result. Individuals only hold it over their contributions. This means that if Sun does relicense, as they are can certainly do, then without the rest of the code (and the closed parts) your contribution will likely become absolutely meaningless to you or anyone else.
Again, it’s an issue of trust.
…that the CDDL is some sort of Sword of Damocles that Sun is forcing contributors to hold over their heads. If that really was the case, can you think of any quicker way of killing the OpenSolaris community?
That’s just a question answering a question that doesn’t really give an answer to what Sun is doing. Because of these question marks there isn’t any real OpenSolaris community to kill, outside of those contributors that Sun finances.
What would Sun gain from such a course of action?
They can have an apparent open source community that they can then walk away from completely at a later date.
Many open source projects are typically not the owners of copyrights of contributions to them, such as the Linux kernel. In the case of Apache and FSF projects, they are two non-profit organisations with many vested interests in them. Sun is a for-profit company with a large amount of proprietary and closed software you would effectively be contributing to, for nothing.
Keep in mind also that Sun holds copyright (albeit jointly) over all of the OpenSolaris codebase as a result. Individuals only hold it over their contributions. This means that if Sun does relicense, as they are can certainly do, then without the rest of the code (and the closed parts) your contribution will likely become absolutely meaningless to you or anyone else.
This is all true, and I think anyone currently going in to the relationship with Sun on this knows it up front and are doing it with there eyes open.
I just want to clarify, though, that the contributors doing so are doing of their free will because of their desire to work with Sun and Solaris, not because anything in the actual license on the code is forcing them too. It’s an attribute of the current community around the OpenSolaris source base, not the source base itself.
If IBM wanted to really play with Sun, they’d fork OpenSolaris and start fixing bugs in a community that does not require shared copyright (or if it does, it goes to a non-profit foundation or whatever). EclipseOS, for lack of a better word.
It would be a waste of time, pety, and spiteful, but still a very annoying thorn in Suns side if IBM were able to build its own, distinct OpenSolaris community outside of what Sun is doing, particularly since the new community could easily take CDDL’d patches and changes from the current OpenSolaris community, but the OpenSolaris community could not easily incorporate changes back.
Which is very, very flimsy indeed. The FSF or some other non-profit organisation is not engaged in a patent license agreement with people like Microsoft. What it does eliminate is the need for Sun to notify other contributors if someone else has a claim over the code used. It’s simply a matter of trust really, and not something anyone can go for unless the cards are on the table.
…
Again, it’s an issue of trust.
…
They can have an apparent open source community that they can then walk away from completely at a later date.
OK, I think I understand the nub of your concerns – you don’t trust Sun. We could have a discussion about the relative merits of the CDDL versus other OSS licenses, but I’d wager that even if Sun released Solaris under the GPL you’d still be suspicious – your concerns are not so much about licensing or copyright issues per se, they are primarily about why Sun is doing this and if they can be trusted.
I’m not privy to any inner secrets (I’m just a grunt) but I’ve been doing Open Source stuff for a while. I was one of the first people to start working on the OpenSolaris program, and I’m now helping out with the process of releasing Java as Open Source. I wouldn’t be involved if I thought there was any malign intent (or even a hint of one) behind Sun embracing Open Source. Sure Sun is not a charity – they hope to make money by doing it (see http://blogs.sun.com/roller/page/jonathan/20050616) – to paraphrase, “It’s all about volume” but I don’t see how that translates into Sun being untrustworthy.
OK, I think I understand the nub of your concerns – you don’t trust Sun. We could have a discussion about the relative merits of the CDDL versus other OSS licenses, but I’d wager that even if Sun released Solaris under the GPL you’d still be suspicious – your concerns are not so much about licensing or copyright issues per se, they are primarily about why Sun is doing this and if they can be trusted.
Based on that, I think GPLing it would alleviate the requirement for trust, because once they’ve taken that step they’ve effectively handed over control. Of course they can close it off at a later date, but they’re unable to use contributed code as a basis to build a closed-source version. Whether you’re a FSF zealot or a pragmatic OSS-supporting organization, you don’t need to worry about your code contributions being one day being used for closed and proprietary gain.
For all the warts and hassles of the GPL, there certainly are many, that is it’s biggest advantage. No trust required. Of course the drawback to GPLing the project in the first place is ultimately losing control, which is no doubt Sun’s biggest concern whether talking about openSolaris or Java et al. That’s absolutely their perogative, but the license ultimately dictates how the project is viewed by potential developers and contributors.
Trust is inevitably the issue, but I don’t think it’s specific to Sun. People would be having this same debate and bringing up the same points if IBM release AIX or HP released HPUX under a CDDL-type license.
*sigh*
Based on that, I think GPLing it would alleviate the requirement for trust, because once they’ve taken that step they’ve effectively handed over control. Of course they can close it off at a later date, but they’re unable to use contributed code as a basis to build a closed-source version. Whether you’re a FSF zealot or a pragmatic OSS-supporting organization, you don’t need to worry about your code contributions being one day being used for closed and proprietary gain.
The license has NOTHING to do with this. CDDL is essentially identical to the GPL regarding distribution of changes to work, it’s significant difference is its file basis vs GPL’s viral basis.
If Sun had GPL’d OpenSolaris, and everything else was the same, specifically the structure of their community, then the same issues arise.
The issue is simply copyright, not license.
If you have copyright, the license is irrelevant as you can change the license at will.
Sun CAN NOT “take back” Open Solaris. Just like Borland/Inprise can not take back Firebird/Interbase. It’s out, it’s free, there’s no turning back.
If Sun shut down the OpenSolaris servers tomorrow, someone would have the sources posted someplace the next day and nothing would change save disruption to the community.
Sun can only incorporate and “take” into closed projects code that is WILLINGLY contributed to them. There’s no subterfuge here. No shenanigans.
Just like the GPL, if you make changes to CDDL based code, you are obliged to make the changes available to those whom you’ve given binaries. Open Solaris is more like the GPL here than the BSD or Apache licenses.
With BSDesque licenses, anyone can take your code in to a dark project and you’ll never see it again. With the CDDL, they can not do that unless you give them copyright to the code.
To play with the OpenSolaris code base on opensolaris.org, at the moment you need to share copyright with Sun. Feel free to copy the code, make your changes however you see fit and distribute them as you like without giving Sun as much as a “By your leave”. There is nothing stopping you, or anyone else, from doing this.
To play with the OpenSolaris code base on opensolaris.org, at the moment you need to share copyright with Sun. Feel free to copy the code, make your changes however you see fit and distribute them as you like without giving Sun as much as a “By your leave”. There is nothing stopping you, or anyone else, from doing this.
That’s the point I’m getting at. So in a nutshell, the alternative to assigning copyright to Sun for contributions to “OpenSolaris” is to fork it and go your own way. That might make for a good community play, but it does not make for a collaborative business model with third-party players.
In simplistic terms I’m arguing that IBM would never have contributed to linux if they had to hypothetically assign copyright to HP. And linux would never have reached it’s current point of universality if IBM and HP both maintained their own seperate versions of the kernel and userland tools based on their own implemented patches and code improvements because they wouldn’t assign copyrights to their competitor. Sure the community has access to code and can modify it as they see fit either way, but how well does that work for ISVs or other related vendors? It’s hard enough getting them to deal with distros that simply package things differently. It would simply be the old days of unix vs unix all over again.
I’m not arguing about the technicalities of OpenSolaris licensing. I’m simply questioning the practicality. Is opening Solaris simply a move to appeal to the FSF community, or are they trying to actually leverage the OSS model for development in the same manner as linux, for example?
For all the warts and hassles of the GPL, there certainly are many, that is it’s biggest advantage. No trust required. Of course the drawback to GPLing the project in the first place is ultimately losing control, which is no doubt Sun’s biggest concern whether talking about openSolaris or Java et al. That’s absolutely their perogative, but the license ultimately dictates how the project is viewed by potential developers and contributors.
How do you lose control of the project just because you use GPL. Has Sun lost control of its UltraSparc T1?
Sun has used GPL in several places when it it the appropriate licence to use.
Alan, a small nit: the ASF does not do copyright assignment, they ask for a written agreement to redistribute contributions under ASL2.0. But the copyright remains at the contibutor alone.
There is a huge value in having copyright assignment on a project, as anyone who’s gone through a cumbersome and long relicensing process knows.
Thanks Dalibor, I stand corrected 🙂
If IBM chose to support and distribute code improvements to OpenSolaris, the CDDL would require them to assign copyright to Sun and give Sun the freedom to ultimately do whatever they wanted down the road, including revising the license and closing off the code.
Any sane project at this point requires copyright attribution, even the Free Software Foundation does. It’s unfair to say that any other organisation that you assign copyright attribution too would not do anything evil, but claim SUN would. Any of them could do sometihng evil like that.
So to IBM’s point, despite the blustering, OpenSolaris isn’t really an open project in the sense of linux and doesn’t encourage outside development beyond a community of outside developers that may not be as concerned with the licensing any more than they would be with a BSD project.
Yes, it is an open project and you have no proof otherwise. Only speculation.
Actually, the FSF couldn’t do that, as their copyright assignment agreement they have with each of their contributors disallows it. They’d be violating thousands of contracts with their contributors if they ever took gcc proprietary, for example.
Actually, the FSF couldn’t do that, as their copyright assignment agreement they have with each of their contributors disallows it. They’d be violating thousands of contracts with their contributors if they ever took gcc proprietary, for example.
Evil is in the eye of the beholder. For example, many of the people that assigned them copyright in the GPLv2 era might be horrified at some of the things that the GPLv3 contains. Yet, the FSF can freely relicense their code under that or any other open source license they release if I’m not mistaken.
I said evil, not proprietary, and the definition of evil is subjective.
Any sane project at this point requires copyright attribution, even the Free Software Foundation does. It’s unfair to say that any other organisation that you assign copyright attribution too would not do anything evil, but claim SUN would. Any of them could do sometihng evil like that.
The difference is Sun competes with IBM, the FSF and any number of OSS projects don’t. And as was pointed out in a different reply, the FSF contractually prohibits themselves from changing the licenses as part of that copyright assignment.
Yes, it is an open project and you have no proof otherwise. Only speculation.
I never said it wasn’t open, I even gave Sun faint props for opening it.
What I did say is that it’s not open in the same sense as linux. And by that I mean that the licensing won’t encourage significant third-party investment in helping develop and accelerate the project.
And you’re right, maybe that’s speculation, but how long will we have to wait to see if IBM, SGI, HP, or any number of otherwise competitive profit-driven companies are willing to contribute time, energy, resources and code to the project before we can deduce it as proof that the license is inhibitive in a manner that linux isn’t?
And to my original post, I’m not knocking CDDL. Sun has every right to release code under any license they choose. My point was that, similar to BSD, it doesn’t encourage collaborative support from third-party companies, which is why linus chose the GPL (due to it’s function, not it’s philosophy) and why linux was readily embraced where scores of non-GPL but still open OSI-approved OSS projects were not. And maybe that’s fine, and Sun isn’t expecting that and will happily continue to do the heavy lifting. But coming back to the original point, this is why IBM will not be embracing openSolaris and does not deem it a truly open project.
I like OpenSolaris and I use OpenSolaris. I personally think this is sour grapes from IBM. OpenSolaris does have a community and Sun tries to be as transparent as they possibly can with OpenSolaris.
The OpenSolaris community does lag behind the BSD and Linux communities in terms of contributions and community members sure, but its only been a year and I think they are making tremendous progress.
Some of us may be underestimating the importance of having an alternative to a single path. The choice of OS and/or the diversity of processors and systems produces far better results in collaboration and competition than any alternative. This is a discussion about options, remember?
1. IBM makes more money per processor from the Power line, but the volume and the related software business based upon Intel products is still greater. IBM just won an award at LinuxWorld for virtualization. They are aware of the issues and want “Power Everywhere.”
2. The CDDL represents careful thought and was an evolutionary step ahead when it was published. In practice, it seems pretty solid to us. “Governance” is an important issue that is addressed in different ways in the context of well-thought-out license agreements.
3. Sun Research Labs will release their Power work to the OpenSolaris community. We are setting up an OpenSolaris project page now for the Power port on http://www.opensolaris.org. Here is some background on the matter:
http://research.sun.com/spotlight/2006/2006-06-14-SolarisPPC.html
4. Originally, the Community had labeled the port “Polaris.”
http://www.blastware.org/
The Sun legal team expressed concerns about the ‘Polaris’ name because it is easy to confuse it with the ‘Solaris’ trademark. As such, we suggested that the project be therefore named ‘OpenSolaris for the Power Architecture’ – or shorter: OpenSolaris on Power. Whatever the name, it is coming and what will happen afterwards will primarily be a function on the parties interested enough to do something with the code. We will continue to support the process.
5. What will probably surprise the IBM’ers quoted in the article is the fact that OpenSolaris on Power will be supported at http://www.power.org/ . There will be many “right” ways to do many things. Choice is a good. In the meanwhile, stop by and check out Power.org. We are giving away t-shirts with the new logo…;-)
http://www.power.org/motion
R&B