Submitted by This morning, at the OpenSolaris Governing Board (OGB) meeting, the following was proposed and unanimously resolved: “The OGB is keen to promote the uptake and open development of OpenSolaris and to work on behalf of the community with Oracle, as such the OGB needs Oracle to appoint a liaison by August 16, 2010, who has the the authority to talk about the future of OpenSolaris and its interaction with the OpenSolaris community otherwise the OGB will take action at the August 23 meeting to trigger the clause in the OGB charter that will return control of the community to Oracle.”
Sun dug a hole where OpenSolaris will be buried. It is called CDDL. GPL hate and incompatibility destroyed another great technology. When will people learn that GPL is only license which makes your open source project immune to hostile takeover?
BSD license?
GPL license wouldn’t have mattered at all in this case. Sun didn’t manage to get a diverse community of contributors for OpenSolaris and insisted on centralized copyright control. Oracle made the situation worse by not following up on promises to the OpenSolaris community and maintaining stone cold silence in response to the OpenSolaris Governance Board making them effectively useless. When a single vendor has such a strong hold on the project, copyright license is least of their worries. OGB is recognizing the fact that they have become mute powerless spectators and taking the right action.
Ok, that too. Control freakishness of Sun had a play there, but if license was GPL (or at least GPL-compatible, like BSD) then a lot of people would get interested in technology, community would get bigger and Oracle would have two choices: keep OpenSolaris going, or let it go and watch people fork it. GPL forces dictatorship to be benevolent.
So Sun used “phuck you Linux” license and even sent their evangelists to talk how CDDL is “more Free” than GPL. That created bad faith and lot of people refused even to look at OpenSolaris code. For example, look what Linux kernel developers said when OpenSolaris was opensourced. Nobody was going to look at it.
Another example, Debian guys thought about making Debian GNU/sunos – SunOS kernel with GNU user-land – but when they read CDDL, they decided to go with FreeBSD kernel and make Debian GNU/kFreeBSD port. If Debian GNU/Sunos was created, that would lead to lots of bug-fixes and QA for sunos kernel.
You are asserting that if OpenSolaris was released under GPL by Sun, it would have automatically gained a diverse community but that fails in the face of reality. Take a look at MySQL which is a project under the GPL license managed by Sun which had similar issues around lack of community and worries about control of Oracle during the Sun acquisition. License alone does not guarantee any community at all. It might have helped or hindered. One cannot say for sure especially since Sun had to mix proprietary code it had licensed from third party folks in OpenSolaris which has yet to be replaced.
Debian’s reading of CDDL is considered incorrect by FSF which considers certain components to fall under the system exception of GPL and hence not incompatible for the purposes of a port. So that is not a compelling example of a problem.
I’m with you about GPL is not the panacea. Anyway I think is a better option than CDDL.
Look at Mysql, it was possible to create the fork MariaDB.
In the case of MariaDB, that is specifically funded by a commercial company and led by a co-founder of MySQL.
If all you want is a fork, CDDL permits that as well. There is nothing preventing OpenSOlaris folks and several such derivatives already exist.
The problem with OpenSolaris is not that the license does not permit folks but that OpenSolaris is almost entirely developed by Sun engineers who now answer to Oracle. If you fork, you wouldn’t be able to do much development because there is not much of a development community outside of Oracle capable of developing a fork of OpenSolaris. GPL license cannot solve that problem.
No, the problem with open solaris is that no one uses it, because it isn’t any better enough than whatever anyone is already using for that task to justify a change.
Which is the problem many alternative OS’s face in this day and age of wintel machines with significant uptimes, and SMP Linux beasts. If we are going to swich to something and make it cool, lets make it really cool. Like, a plan 9 clone with an api written entirely in haskell.
I use it at home for ZFS for a home fileserver and as a MySQL server. I like ZFS so much that although I’ve contemplated switching to Debian tomake it easier to set up other services, I just can’t bear the thought of losing ZFS’s flexibility. And Linux RAID/LVM (which I use on my other machines) does the trick, it still comes nowhere near the partition flexibility of ZFS.
I’ve heard some rumours that ZFS may be coming to Linux though — depending on how that goes I’d consider switching, though I didn’t find FreeBSD or Nexenta to be a suitable replacement for OpenSolaris either.
You are right, of course. Open solaris has a special advantage in the storage space niche due to ZFS. People do use it there. I guess that is a spot where they have an advantage due to the lack of GPL.
I still feel the larger marketplace for x86 based *nix servers is dominated by Linux distros. The lack of a wider adoption of Open solaris, is its main problem and its biggest obstacle to future development by non oracle developers.
Plus, there are many indications that BTRFS will soon mature, and the ZFS advantage will fade away.
MySQL have community, and if Oracle tries to kill MySQL, forks will take over. There are already few forks like MariaDB, Drizzle, XtraDB… but nobody takes notice of those because Oracle is currently behaving ok and not trying to kill MySQL. As I said, GPL makes dictatorship to be benevolent, and if dictator is harmful to the project, then project will fork away.
Also, I don’t think that copyright assignment policy is bad by design, but some companies are abusing it.
As for Debian interpretation of CDDL..? What issue are you implying? One with cdrtools, or one with sunos port?
Copyright assignment policies allow centralized copyright control and even though Oracle does release MySQL under the GPL, the license does not apply to Oracle and Oracle is free to sell proprietary versions.
If you want forks, OpenSolaris has several derivatives and forks that already exist. Again, CDDL does not prevent that at all.
As far as Debian is concerned, it has no lawyers and interpretation of license is done by developers which is problematic and in the case of porting, it was not prevented by CDDL. Debian might have thought so but does not make it correct.
Ok, you are dismissing Debian because they are no lawyers so they know nothing. IIRC, Debian issues with CDDL were:
-choice of venue clause, fixed at Santa Clara for OpenSolaris, but changeable for other projects written by others. That in fact creates a separate versions of CDDL, for example Sun’s one is with Santa Clara, but some dude would chose his district and effectively make it incompatible with Sun’s version. Only way to get around of this issue would be to force everyone into choosing Santa Clara venue, same as Sun. Of course, this was problem for Sun because they ask for copyright assignment so that they not need to worry about that. But it could pose great problem for Debian if they allowed themselves to be dependent on CDDL code, because some (most?) of contributors are not comfortable with assigning copyrights to public company, and don’t live in Santa Clara so they can be conformable with choosing it.
-choice of law clause, more or less same issue as above. When you take code from “CDDL Santa Clara” source file, you can’t put it into “CDDL Mexico” source file, you can only link it. But that would create legal mess for project like Debian which like you said, has no lawyers.
-patent clause of CDDL is to narrow and only protects corporations with patents. If you have no patents to strike back, it only protects “Original Software”, effectively meaning that if Oracle stops releasing OpenSolaris, forks will not get patent grant. Also, patent grant ceases if certain source files are removed. Read CDDL and look at definition of “Original Software”, and license grants wrt section 2.
About forks, well did you know that NetApp is chasing ZFS storage start-ups for patent infringement? You know why? Because they are cutting deal with Larry. NetApp exterminates ZFS startups, and Larry position (newly re-branded) ZFS Storage appliance in a way that it doesn’t threaten NetApp’s proprietary WAFL. That means they stall open source development, and kill it of at the end, continuing only proprietary branch of Solaris. If FreeBSD (or FreeNAS) moves to threaten WAFL with ZFS storage, it could get sued too.
Edited 2010-07-14 07:28 UTC
No I am dismissing them because they are blown up issues. None of these problems are specific to CDDL and while choice of venue clauses are problematic they dont have a license non-free by itself and does not prevent porting. GPL does not prevent patent infringement lawsuits either.
There’s plenty of GPL OSs out there that are struggling.
Simply stating that a GPL licence would attract a community is somewhat deluded. Particularly when all that would likely happen would be the popular technologies in OpenSolaris (ZFS, dtrace, etc) would have been gutted and implemented in Linux.
So as things stand, the CDDL licence has potentially brought people to Solaris who were interested in technology and weren’t loyal to a platform. And those that refused to move because of loyalties wouldn’t have moved anyway.
But those kind of developers wouldn’t have looked at OpenSolaris anyway.
Linux fanboys will stick with Linux. Sun knew this, so they wanted to protect their assets rather than lose their biggest bargaining chips to a developers that are unlikely to switch platforms regardless.
A GNU/SunOS does exist. In fact NexentaOS has been around for a few years now and have been well reported on technology sites like this.
And yet, that’s exactly what the Nexenta folks did. Took the OSol kernel, put the Ubuntu userland on top, and released it to the public.
Just because one group of devs decided it wasn’t worth the effort doesn’t mean the entire world did.
They advertise like they did it, but they in fact didn’t.
You see, Nexenta is in fact OpenSolaris with GNOME and dpkg. It still uses sun libc as oposed to planed Debian port which would use glibc and all GNU userland except Linux kernel, which would be replaced with SunOS. Thet would help in getting more potable glibc but SunOS kernel would get lot of benefits too. If fact, Solaris could use glibc because is superior library. It would be real gain for both. Note that debian people are contributing both to Linux and now BSD kernels, they would do it for SunOS too.
This way, Nexenta just uses OpenSolaris and repackages things to .deb. They do no kernel hacking just repackage things. They are no use for Oracle as they are not contributing, just leaching. No wonder Larry wants them dead.
Go look through SourceForge projects if you think that the GPL will magically attract developers. It’s mostly a graveyard.
I think a bigger problem with OpenSolaris is that they never made a good case for switching from Linux or FreeBSD. Linux has better hardware support, FreeBSD has ZFS and Dtrace. Why take the time to move your servers to OpenSolaris? There was never a good answer.
Even so, OpenSolaris was still hugely reliant on Sun for key parts of the operating system which ‘Project Emancipation’ was meant to ’emancipate’ OpenSolaris from Sun but never went anywhere. OpenSolaris is still reliant on Oracle and until those closed source binaries are fully replaced OpenSolaris will never be truly free. It is disappointing that the opening of Solaris started with such a hiss and a roar – here it is now stuck in a quagmire.
In all honesty it is absolutely pathetic, the last post to emancipation mailing list was December 2009 – it really goes to show that rather than Sun actually ever having their heart into transforming Solaris into a truly open source community they seem to have turned the internet as a dumping ground for a product they seem to have lost interest in developing years ago. It is really depressing that something I saw as a possible alternative to the Linux hegemony has crashed and burned so quickly.
I can see OpenSolaris as still a bit useful, even if it didn’t develop a community of developers outside Oracle. Assuming OpenSolaris is still the basis for Solaris 11, it’s sort of like a big open beta test. The changes of CDE to Gnome, adption of a modern package system with online repositories, and switching to X.org for graphics are major changes that update Solaris. Solaris is the most approachable and easiest to use proprietary Unix available. There’s no crazy hardware to find and getting discs of the OS is easy. I’d imagine more non-enterprise programs work on Solaris than HP/UX or AIX. It’s pretty approachable.
It’s unfortunate OpenSolaris never got a vibrant developer community outside Oracle, and is still dependent. On the other hand, I can’t think that Sun/Oracle lost much making Solaris open source without one. Oracle can still say, “here, see the latest work we’re doing, play with all the technology we offer, and if you’re interested, ask us about our hardware and services that take full advantage . . . blah blah . . synergy . . . whatever.” Seems like it could pay off, and it wouldn’t be expensive or difficult. Then again, Oracle might not work that way, and Oracle might not care.
The two biggest flaws of OpenSolaris for me were always the following:
1) The use of HAL in lieu of simply biting the bullet and adding native Solaris support to GNOME. This is a symptom of a larger problem, when ever they add something it is always half baked – BlueTooth support is a half baked port of the NetBSD stack for example. Half finished half baked features being added and once added nothing be done to further enhance them. So they sell servers – great, but what about all those developers, admins and so forth who could be potential customers and advocates of OpenSolaris if they could only run it and provided with a reasonable experience?
2) Crappy hardware support and quite frankly there is no excuse for that. In the case of Linux adding hardware support is a little more complex when you’re dealing with a kernel with licence issues that need to be taken into consideration and no stable driver API. On the other hand OpenSolaris has a large corporate backing it, that corporation also sells hardware *AND* that corporation has links with other hardware suppliers! it should be relatively easy to add hardware support, even if it is providing closed source binary driver downloads.
I truly believe if they got those two issues sorted out there would be nothing stopping OpenSolaris from being widely adopted from laptop to desktop to workstation. What is holding OpenSolaris back isn’t major fundamental flaws in the operating system but very basic things like hardware support that can be addressed relatively easy if the parent company is willing to allocate the resources to do so.
If it were me and I was grand poobah Oracle, I would hire another 2,000 programmers, put them all on a 3 year contract and get them doing nothing but sleeping, eating and breathing drivers – every waking moment of their existence either writing drivers or getting the specifications to write those drivers. Drivers, drivers and more drivers – just when they though they’ve covered all the bases, get them writing more drivers. Once you lay that crucial foundation then it is easy to top up additional drivers without needing to have a large staff – but you need to get OpenSolaris up to the point where you can pick up a OpenSolaris iso, throw it on a cd, dump it into a random computer and find that all the hardware in the computer itself works.
Edited 2010-07-15 12:07 UTC
This kind of crazy hippie talk makes baby unicorns cry. It’s nonsense really.
The fact is that Open Solaris was an corporate experiment that failed. Open sourcing Solaris failed to create enough paying users to make it worth the effort. it did not even appreciably increase market share. These are probably concepts that you are not familiar with: profits and market share. Look them up on wiki. If Solaris was open sourced with GPL3 (God forbid even the idea!) it would not have changed history.
Oracle is now taking a much less tolerant attitude. All these unprofitable Sun projects will now come to an end. They serve no purpose for Larry Ellison’s shareholders. Open Solaris will now be rolled up with little fanfare. That is all. Good day.
You are not familiar with fact that open source can generate profits, it does so for Red Hat and many others. Even for Oracle.
If Sun decided to opensource Solaris, they should have done it properly, or not do it. OpenSolaris was failed experiment because they didn’t do it properly. And even though they didn’t do it properly, even though Sun didn’t take advantage of all benefits of open source (they took advantage of free redistribution and bug reports by non paying users, but didn’t expand development community and accelerated development) , OpenSolaris managed to stop decline of Solaris for certain period of time. At least until financial crisis hit.
If they didn’t open source it, it would be dead OS by now, and Sun would be even cheaper buy for Oracle.
If they opensourced it under GPLv3, it would certainly helped to reduce development costs and increase pace of innovation. But it might not have helped to save company because Sun lost money on hardware, and software support revenue couldn’t supplant that. You could argue that writing on the wall was since year 2000. But at least in that case OpenSolaris would live on.
This way, both Sun and OpenSolaris are dead.
How does the gpl increase the pace of innovation..?
And if it were the case why would we see gpl licensed “copies” of Sun technology?
I totally agree that they have should done it properly or not at all. The way things are, OpenSolaris should drop the “Open” from its name. It’s a shame.
CDDL, GPL, BSD… all licenses would work if Sun really had the intention of creating a healthy open source project but they want to hold their feet in the proprietary world and created this mess.
How is it done improperby?
No, under normal circumstances it can’t. Every company can’t be Red Hat and package a software, that was written mostly by someone else. I see you’re a gnu fundamentalist and you won’t believe me, so just think about what percentage of successful software is open source. Oracle DB? WebSphere? DB2? Anything from Microsoft?
Yes, they shouldn’t have done it, because it made them fail in the market. And they shouldn’t have made incompetent Jonathan a CEO in the first place.
Had to fix that one for you.
What would really happen is, that Linux would reap all the great functionality and leave it out in the cold, just like they did to the BSDs.
It didn’t help when Sun Sales wouldn’t return calls of companies who had been running OpenSolaris and wanted to buy license and support.
Sun Sales ignored these potential customers because they weren’t considered big enough to matter.
Leaving money on the table was clearly a sign that Sun was going down.
We need an Oracle person to tell us what to do, else we’re going to give Oracle complete control of the project? Am I interpreting that right? Because if I am, I must say, I don’t understand the logic behind it. Seems to me, given an ultimatum like that, best thing for Oracle is to just do nothing.
Oracle already has effectively complete control over the project. Dissolution of the OGB would just make the control more publicly visible.
Yes, I know, so what’s the point of the OGB’s ultimatum? Do nothing, Oracle wins. Appoint an Oracle boss, Oracle wins. Why the complicated dance around the issue?
OGB is just following a procedure as outlined in the constitution. OGB doesn’t want to appoint a Oracle boss. It wants a representative from Oracle to answer the questions from the community instead of being silent. If Oracle fails to get a representative, OGB is ineffective and failed its mandate and hence dissolving itself.
This is not about winning. Oracle is the copyright holder, they have won by default (if we are talking about OpenSolaris, not a fork). The OGB is merely acknowledging it makes no sense to hold an official title that means nothing.
I think they are basically saying that is Oracle won’t deal with them, then the board will fold and oracle will be given responsibility for the OpenSolaris project.
Right now, Oracle can hide behind the group and wash their hands of it. With the board dissolved Oracle can’t put it off any more. The idea being, I think, Oracle will have to either keep the project alive or kill it and take full responsibility for the result.
more open is more the future. a future full of freebsd is very more far away, but they are happily puttering toward that future, copying everything of value in opensolaris.
Can we get a more biazed and trolled headline please? I don’t think you guys are trying enougth.
gonna need that stapler too.
Seriously though who didn’t see this coming? Oh that’s right all the deluded OpenSolaris fans.
OpenSolaris just doesn’t fit in very well when Oracle already has unbreakable Linux and Solaris. It isn’t needed and it never attracted a healthy community.