Weekly Edition
Return to the Announcements pageRecent Features
| |||||||||||||
How Sun's need to control the code cost them the company (ZDNet)
Jeremy Allison looks back at the demise of Sun Microsystems on ZDNet. "The Solaris operating system, the Java language and virtual machine, the OpenOffice office suite - all of the really large software projects that Sun released - had strings attached that stopped any real external community from forming around the code. Usually it was the demand that any code contributions be contributed directly to Sun for their own use in proprietary products that was the major failing of all the Sun 'community' projects. Poor licensing choices, demands for ownership of all contributors work, ignoring contributors outside of Sun, all of these can be blamed for Suns inability to maintain active coding communities around their Open Source code, but in the end it comes down to the desire to maintain control and ownership of the code at all costs. People are smart enough to understand when theyre being taken advantage of, especially programmers."
(Log in to post comments)
How Sun's need to control the code cost them the company (ZDNet) Posted Mar 2, 2010 17:42 UTC (Tue) by donbarry (guest, #10485) [Link]
This sort of "what's mine is mine, what's yours is negotiable" inflicts
a variety of communities when greed, bureaucratic conservatism, or even simple fear come to dominate the relationship. "No one has ever been fired for keeping the code" could be their mantra (and of course it's untrue).
Programmers are smart. When TWiki went from a community project to a
Its hard to find this degree of solidarity in ordinary labor communities.
How Sun's need to control the code cost them the company (ZDNet) Posted Mar 3, 2010 2:52 UTC (Wed) by pabs (subscriber, #43278) [Link]
Another organisation like TWiki is the ntfs-3g project. I don't see any fork of that though, I guess there was no development community before they did that.
Heh Posted Mar 2, 2010 17:46 UTC (Tue) by robilad (guest, #27163) [Link]
"The Solaris operating system, the Java language and virtual machine, the OpenOffice office suite - all of the really large software projects that Sun released - had strings attached that stopped any real external community from forming around the code."
Out of around 180 individuals with commit access to the various OpenJDK project, around 60 of them are 'external', assuming I interpret the amusingly bizarre construct of a 'real external community' here correctly.
Heh Posted Mar 2, 2010 18:13 UTC (Tue) by rahulsundaram (subscriber, #21946) [Link]
External in what sense? Are they Sun employees and are they required sign
copyright agreements? How much code are this "external" people contributing
Heh Posted Mar 2, 2010 18:30 UTC (Tue) by robilad (guest, #27163) [Link]
"External in what sense?"
You'll have to ask Jeremy what he means.
"Are they Sun employees"
Around 60 of the around 180 committers on OpenJDK projects are not Sun or Oracle employees.
"and are they required sign copyright agreements?
Yes, of course. This is all quite trivial to find out on the OpenJDK web site, fwiw - go to http://openjdk.java.net and enjoy yourself.
"How much code are this "external" people contributing"
Since OpenJDK is a free software project, you can hg fclone all the different forests and find out for yourself - there are 26 projects within OpenJDK, 12 of them being led by developers that don't work for Sun or Oracle.
Heh Posted Mar 2, 2010 18:56 UTC (Tue) by rahulsundaram (subscriber, #21946) [Link]
Copyright assignment means Sun wanted to keep all the control and that is
one of the points that he mentions and it is always a pain point in building a true community and raw count of non-Sun contributors alone doesn't offer much of a counter point
Heh Posted Mar 2, 2010 19:00 UTC (Tue) by robilad (guest, #27163) [Link]
When you don't have the facts on your side, you can always argue semantics of 'true community', indeed. ;)
Heh Posted Mar 2, 2010 21:30 UTC (Tue) by rahulsundaram (subscriber, #21946) [Link]
Smiley doesn't hide the facts that Sun's demand and insistence of copyright
assignment has hampered the projects it leads and this includes many many projects such as OpenOffice and this has been pointed out in meticulous amount of detail by many
http://people.gnome.org/~michael/blog/ooo-commit-stats-20...
If you want to present facts you can provide the data in a similar fashion
Heh Posted Mar 3, 2010 9:31 UTC (Wed) by trasz (guest, #45786) [Link]
IIRC, FSF requires exactly the same copyright assignment for its own projects (GCC et al).
Companies try to bank on the goodwill created by the FSF copyright assignment process Posted Mar 3, 2010 9:46 UTC (Wed) by mjw (subscriber, #16740) [Link]
Companies try to bank on the goodwill created by the FSF copyright assignment process Posted Mar 3, 2010 10:45 UTC (Wed) by trasz (guest, #45786) [Link]
From your point of view - perhaps. From my point of view, they are pretty much the same thing - they require me to pass my copyright to some other party.
Companies try to bank on the goodwill created by the FSF copyright assignment process Posted Mar 3, 2010 10:53 UTC (Wed) by mjw (subscriber, #16740) [Link]
You might want to read LWN's analysis then about "Community contributions and copyright assignment" http://lwn.net/Articles/359013/
The difference as that article points out is that "Agreements like Sun's and SugarCRM's are common when dealing with corporate-owned projects; clearly prioritize control and the ability to take things proprietary over the creation of an independent development community. More community-oriented projects, instead, tend to take a different approach to contributor agreements."
Companies try to bank on the goodwill created by the FSF copyright assignment process Posted Mar 3, 2010 13:14 UTC (Wed) by nix (subscriber, #2304) [Link]
So in one comment you've jumped from 'exactly the same' to 'pretty much
the same thing', while claiming your position has not shifted. Interesting.
Actually they are mostly the same... Posted Mar 3, 2010 22:16 UTC (Wed) by khim (subscriber, #9252) [Link] FSF also requires copyright assignments and this led to troubles with forks and lack of community support many times (Emacs, GCC/EGCS, etc). The difference here is clear: FSF is non-profit organization and such troubles just mean it'll have some trouble and some of it's projects will fail (see "HURD vs Linux", "DotGNU vs Mono", etc) nothing else, but Sun is commercial company and must pay the bills... Also "exploitation angle" is much weaker with FSF: it's assignment does not give the FSF right to sell closed-source forks.
Actually they are mostly the same... Posted Mar 3, 2010 22:48 UTC (Wed) by nix (subscriber, #2304) [Link]
The GCC/EGCS fork had nothing to do with copyright assignment. Why do
people keep saying this?
It was a dispute over development methodology, nothing more. Copyright
True Community Posted Mar 2, 2010 19:10 UTC (Tue) by mjw (subscriber, #16740) [Link]
You can always just contribute to the IcedTea project which is a friendly fork of OpenJDK that doesn't require you sign over all right to Sun/Oracle if you don't want to: http://icedtea.classpath.org/
True Community Posted Mar 2, 2010 21:23 UTC (Tue) by tzafrir (subscriber, #11501) [Link]
And likewise http://go-oo.org/ for OO.o . How much effort has been duplicated there?
True Community Posted Mar 2, 2010 21:39 UTC (Tue) by rahulsundaram (subscriber, #21946) [Link]
Yes and the need for such "friendly" forks before a community can form
should have been a big warning sign for Sun and even though it was later partially resolved this blog from you a while back showed what this control could lead to
http://gnu.wildebeest.org/diary/2009/11/14/trusting-compa...
Copyright assigmnent Posted Mar 2, 2010 21:30 UTC (Tue) by lkundrak (subscriber, #43452) [Link] Copyright assignment means Sun wanted to keep all the controlSince when does copyright assignment imply complete control? What's important here is the license agreement. Once you release the code under a free license you can't close the code you've written no matter who owns the copyright.
Copyright assigmnent Posted Mar 2, 2010 21:42 UTC (Tue) by rahulsundaram (subscriber, #21946) [Link]
I was referring to control from the development perspective obviously
As MySQL controversies have amply demonstrated to us one entity having all
Copyright assigmnent Posted Mar 5, 2010 9:41 UTC (Fri) by sitaram (subscriber, #5959) [Link]
copyright assignment allows them to change the license afterward.
Sure you can always fork the last GPL version, and that's fine in theory. In practice it splits developer and user mindshare, and should really be a last resort.
And to the folks comparing this to FSF requiring copyright assignment, while I'm not a great fan of everything RMS says, I do trust that the FSF will not turn something proprietary tomorrow :)
Copyright assigmnent Posted Mar 5, 2010 10:21 UTC (Fri) by rahulsundaram (subscriber, #21946) [Link]
Actually FSF cannot since they counter promise in their copyright agreement
to always retain the code as Free software
http://www.ebb.org/bkuhn/blog/2010/02/01/copyright-not-al...
Java did better ... Posted Mar 3, 2010 7:22 UTC (Wed) by mmeeks (subscriber, #56090) [Link]
Hi Dalibor,
Good to see your Java community leadership goodness on display :-)
I think it is a fair point that Java has done better than the disaster of OpenOffice; presumably by employing better community governance guys, and making less obviously aggressive decisions. Perhaps also because the project is newer, and the abuses of trust and cynicism at "someday, later we'll fix that ..." hasn't mounted up yet.
Ultimately, I am certain that if Java was licensed and distributed in such a way that there was no per-unit market for it, and in such a way that Sun could accept the (inbound) license terms itself - there would be no need for copyright assignment, the code could become genuinely independent, and of course Java would become more ubiquitous.
ATB.
60 out of 180, and how statistics can lie Posted Mar 3, 2010 23:37 UTC (Wed) by coriordan (guest, #7544) [Link]
Statistics of commit access prove very little. Those 60 might be mostly made up of inactive people, or people with low activity, or working on experimental things that aren't being shipped, etc. etc. Project leader statistics are the same - which projects? the libc, or some experimental thingy?
> you can hg fclone all the different forests and find out for yourself[...]
I don't even have "hg" installed and I don't know what a "forest" is.
I haven't said you're wrong - just that those means of proof are weak.
Great write-up ... Posted Mar 3, 2010 7:12 UTC (Wed) by mmeeks (subscriber, #56090) [Link]
Go Jeremy ! an insightful analysis, as usual.
The final outcome is so tragic, given so many really positive moves: opening up code etc. To bungle all this goodness due to an accumulation of silly and bungled details is really, really sad.
Ultimately, if Oracle were to drop copyright assignment, and pick a license for others that they might accept themselves, they might discover they accidentally created a thriving community around the software they had set free. They would also have no need to be scared of 'forks' (which are IMHO rather a good thing, testing new directions - of course, with a common license they can be reconciled after the fact).
How Sun's need to control the code cost them the company (ZDNet) Posted Mar 3, 2010 7:49 UTC (Wed) by rsidd (subscriber, #2582) [Link]
Can it be that the demise of Sun has nothing to do with the software side,
and everything to do with the hardware side of things? Sparc couldn't really keep up with Intel's offerings. Sun eventually started offering AMD opteron systems too, and that worked for a while, but it was still overpriced stuff compared to opteron machines from competitors. Plus they refused to sell windows computers. No proprietary Unix has done well post- Linux, and Solaris probably did better than the rest -- but the firms behind the other proprietary Unices (IBM and HP, principally) had a much more diverse set of offerings, dominated by Intel and Windows.
How Sun's need to control the code cost them the company (ZDNet) Posted Mar 3, 2010 8:40 UTC (Wed) by hppnq (guest, #14462) [Link] Exactly, although I think Sun lost the Unix server battle years ago to HP and IBM.
How Sun's need to control the code cost them the company (ZDNet) Posted Mar 3, 2010 10:07 UTC (Wed) by jmansion (guest, #36515) [Link]
I'm not sure Jeremy is being fair. Sun needed to retain control in order to have a relationship with customers than include commitments to roadmaps. They may have handled community badly, but they have always had an ethic of careful and conservative release management, which also requires an element of control.
Sun is also quite different in trying to free huge and complex code bases that have already benefited from massive investment from their inception. They spent a lot on R&D. Red Hat have certainly never done anything similar: Novell are probably closest in terms of creating something from scratch, given the number of coders on mono related tasks. I hope it works out well for them.
An unfortunate corrolary might be that, actually, it is very hard to justify doing major R&D (ie most of the initial development of a complex system) with a view to open release. Sun tried, and look where it got them: Novell's financial performance is hardly stellar to date either. Projects like gcc and Samba started very small and modestly, and grew - over a long time. They also don't have major R&D spend (except in gcc's case by hardware vendors). Similarly, postgres grew (slowly) out of academic work.
The real question is: can yu create a community around something that exists? Or is growing it organically part of the community building process? And where does that leave projects that must commit to and hit deadlines to align with externals? Major Solaris reelases tended to align to major hardware improvements in sparc enterprise systems. How can vendors ever do that based on an external muddle-along-and-release-when-its-ready system?
The opening of Qt is interesting - but again you're looking at something which isn't the company's primary business - it helps them sell phones and services, much like IBM and others sell hardware and services off Linux, and Google (currently) contribute code to help own internet advertising.
To me this is worrying. It seems hard to create a business around open source unless you give up on primary R&D and focus on marketing and a bit of sexy value add to help win sales - and where's the fun in that? Where will the R&D come from to create non-trivial product? Currently, it comes from people selling hardware, advertising, or an 'assured compatibility' (ie Red Hat - customers want a near-monopoly, so they have more choice in third party assets and support). None of these people do primary R&D with a view to selling software.
It will be very interesting to see how Oracle (and to some extent Novell) work things in the future, and even then Oracle can spend as part of marketing brand awareness and goodwill, and a desire to keep MS and IBM honest on platforms for their DB and app suites.
James
How Sun's need to control the code cost them the company (ZDNet) Posted Mar 4, 2010 16:16 UTC (Thu) by nevyn (subscriber, #33129) [Link]
> Sun is also quite different in trying to free huge and complex code bases
> that have already benefited from massive investment from their inception. > They spent a lot on R&D. Red Hat have certainly never done anything > similar
Red Hat initially did RHN (aka. spacewalk) as proprietary software (with
Also, due to Red Hat's open nature, I think you may be missing the fact
As for Sun's "R&D", they have Zfs and Dtrace. Both fine, I guess, and both
How Sun's need to control the code cost them the company (ZDNet) Posted Mar 3, 2010 11:03 UTC (Wed) by sylware (guest, #35259) [Link] People are smart enough to understand when theyre being taken advantage of, especially programmers.Feels like Linux drivers on a BSD-like license with a kernel abstraction layer (=kludge)...
How Sun's need to control the code cost them the company (ZDNet) Posted Mar 4, 2010 11:39 UTC (Thu) by imcdnzl (guest, #28899) [Link]
I was actually quite astounded by this blog post as I wonder how much of
this is him crying for help within Google given their practices with Android as documented here at lwn - http://tiny.jandi.co.uk/lwnandroid
Posted more on my blog about this at http://tiny.jandi.co.uk/zsnjm
How Sun's need to control the code cost them the company (ZDNet) Posted Mar 4, 2010 13:16 UTC (Thu) by rahulsundaram (subscriber, #21946) [Link]
Your confusion seem to be because of confusing individual opinions with the
company that he works for and the company's actions in a completely unrelated team
How Sun's need to control the code cost them the company (ZDNet) Posted Mar 5, 2010 15:25 UTC (Fri) by cdmiller (subscriber, #2813) [Link]
About 12 years ago I told my uncle (one of Sun's global regional sales managers, now at HP) that SUN needed to pull their heads out and embrace Linux on Sparc. We geeks love SUN hardware I explained, but detest Solaris and all the hoops needed to keep a Solaris system running. Linux kicked butt on Sparc at the time and there was a great Sparc Linux community forming. Since then I watched Java, the Java desktop, the buy of Star Office, Open Solaris, and the buy of Mysql. I also watched SUN hardware slowly suffer in quality. All I could do was watch the disaster unfold. I had already said my piece.
|
|||||||||||||