Slashdot Mirror
Sun Spokesman Says "We Screwed Up On Open Source"
An anonymous reader sends along a video from Builder AU, in which Sun's chief open source officer Simon Phipps describes 2001-2002 as 'a period where Sun 'screwed up' in their dealings with the open source community. Phipps says that Sun is trying to remedy the situation with the open sourcing of Java, Solaris, and the rest of Sun's software."
Thank you Sun for all the great products that you have open sourced. Unlike your competitors, you have outsourced your crown jewels.
Phipps says that Sun is trying to remedy the situation with the open sourcing of Java, Solaris, and the rest of Sun's software."
GPLing ZFS would go a long way with me!
How about the other years? or is 2001-2002 the period they screwed up the worst ?
"A door is what a dog is perpetually on the wrong side of" - Ogden Nash
I am still skeptical of Sun. They are coerced into behaving nicely by the huge open source community which is not that much interested in what they have to offer anymore and have a lot of influence in the market. And let's face it, they opened up Java after IcedTea is out for quite a while now.
http://ebgp.net/ccc/
They restored some respect that they lost from me. Lets see how it goes from here.
To be fair they wanted to open source the code in Q4 1999, however their Java ftp client just finished loading a few months ago so they couldn't upload the code until just now.
you might be asking the wrong company there - as far as i recall, the main jython developer also wrote ironpython, and now works for microsoft who seem to not really take python seriously as its a bit of a bolt-on hack and not nicely integrated into visual studio like c++ etc; they're not exactly the kings of opensource either....
plus, even though the jython library version is out-of-date, it still makes c-python look like a snail - and i never thought i'd say that java is faster than c!
hopefully python3000 will bring us speed if not compatibility.
#include <sig.h>
I appreciate how much effort it must have taken for Sun to move this far on open source. Nevertheless, I think Sun is still screwing up.
Solaris, for example, is being positioned as an alternative to Linux: it's "pick us or pick Linux". From an open source point of view, it would be better if Sun picked a license that allowed the best parts of Solaris and Linux to be combined, and for end users to decide what those best parts are.
For Java, Sun still has most of the control, they have torpedoed attempts to certify Apache-licensed implementations as Java compliant, and their dual licensing scheme for Sun Java means that the project just isn't run the way an open source project ought to be run.
In the short term, Sun's behavior is disruptive for open source, but sadly not in the positive newspeak sense, but in the sense of merely annoying a lot of people for no good reason.
In the long term, Sun is going to lose with Solaris and Java if they persist in their take-it-or-leave-it approach to open source. If they want the technologies to survive in some form, they need to allow a mix-and-match approach; that's what open source is really all about.
So Sun decides that they can't make money by pushing the open source community around, so now they come on their knees and beg forgiveness before they have anymore problems.
/me wraps his arm around Sun's shoulder
Hmm...
It's alright man. Good to have you with us.
Sun has, and has had, some great products in the past, and some of their hardware is still pretty excellent, but the problem with the company is that they still have a deep rooted protectionist attitude towards SPARC and Solaris. Why do you think it took so long to get Solaris on x86, why it took so long for Sun to accept that x86 servers was where the growth was, why most of Sun's customers still get Linux pre-installed on Sun's systems and why Sun paid a couple of billion for an excellent business opportunity in Cobalt, and then promptly destroyed it?
If they could make Solaris and SPARC stand out and pay off then fine, but they can't hence the half-hearted and pretty sad move to 'open source' Solaris just so all their consultants and execs can run around trying to tell us that it's 'just like Linux'. However, in the cold hard light of day, Linux ate Solaris's lunch, and SPARC just competes too closely with x86 based servers without the comparable performance. SPARC is so inferior to x86 in terms of raw performance it's so laughable. Solaris also suffers from the fact that Sun just don't have the resources to push development to where Linux and other operating systems are, and these days it is increasingly expensive to try and maintain an entire OS yourself.
In terms of open source, Sun's problem is that the vast majority of open source software is written for Linux and the BSDs first. No one thinks of Solaris as their first platform of consideration, and it's difficult to see why they should do so now. It's still like that now, and it was still like that a few years ago when a former employer scratched its head trying to work out why Zope and Python performance was so terrible on Solaris and an UltraSPARC. A Sun guy even recompiled Python in Forte. The bottom line answer we got from the Python devs was "We use open source systems, and possibly Windows, first and foremost on x86 systems, x86 and Linux performs better anyway, and while we'd like to help, we just don't care about your corner case problem on an OS and hardware we don't have access to and can't reproduce. Just use Linux and x86". That's not literal, but it's the general gist, and I couldn't say I blamed them.
The solution? They moved to a far cheaper x86 system with Linux, they had no installation problems with Python as it came within the package management system itself, things were far easier to manage, performance increased exponentially which pleased everyone and Python and Zope ran with no issues whatsoever. That still holds true today.
Look at this post here from Groklaw, reviewing the testimony from the SCO v Novell trial.
PJ notes that SCO enacted a license, illegally according to Novell, with Sun in 2003 that allowed Sun to open source Unix Sys V. Knowing they had that, Sun still allowed SCO to embark on their SCOSource campaign against IBM and Linux users for allegedly putting Sys V code into Linux.
[Java]'s a great learning language, like Turing was back in the 80's-90's or so, and it can be used to teach the basics of OOP and OOP compiler design.
Java is a useful language with a well-designed and extensive set of libraries that make it one of the best languages available for implementing most kinds of business applications.
What it isn't, in fact, is a good learning language.
How they EVER planned to make money off java, is beyond me, [well, maybe the embedded versions]. This should have been open-source from the start.
1. The original plan for Java was as a language for embedded systems. The other applications were added as an afterthought, effectively.
2. It's hard to see how they would have been better off had it been free software[1] from the beginning. The language was adopted at a phenomenal rate, has had a huge amount of community input and is now effectively one of the three most important languages for commercial programming today (along with C++ and C#).
Sun's lack of profit from Java stems primarily from the fact that they never developed the kind of leading support software for it that the development community required, leaving things like IDEs and application servers for others to successfully commercialise. If Sun had produced an IDE even a fraction as good as eclipse, or a server environment as robust as IBM WebLogic, they could have made a fortune from it. But they failed to achieve that.
[1]: I use the term to avoid confusion: Java has always been a "source code available" product, but simply did not have the redistributability of modified versions that we expect from free software. Many people understand "open source" to include the former.
Reading this story seem to indicate that 2003 and forward should have been part of the "screw up" period.
It is good to see Sun throw their weight in to support open source.
Carbon based humanoid in training.
Yeah, they screwed up big time. Groklaw has a nice article up on their involvement with SCO:
http://www.groklaw.net/article.php?story=20080625020853732
And they are still screwing up. Just have a look at the legal agreement you need to agree to when downloading even the *specification* from any of the Java Community Process groups where the project lead works for Sun. Evil and completely unacceptable terms for open-source developers...
And in those projects where they have released the source under a free license, they still keep an iron hand on the development process. So unless you work for sun, you need to beg to get your changes in (and sign all sorts of agreements). Closed bug-reporting systems. Version-control repositories that you need to apply to get read-only access to. Closed mailing lists. Design meetings held in person (Sun employees only of course).
This is a company that has a *long* way to go before they understand what Open Source is about.
Or, less charitably, this is a company that does indeed understand what open-source is about and is manipulating the system. Yes, once the source is released a fork is then possible, but for a large project inertia and an existing pool of developers all from one company make that something that takes real anger to do. So the changes Sun has made so far don't achieve a whole lot; they still completely control the direction their open-source projects take.
Real OSS companies are different; they contribute upstream, allow derivatives downstream, and are open in their process. A whole world of difference. See RedHat for a good example.
Sure, Simon Phipps's quote makes a good headline, but between OpenOffice.org, Netbeans, Glassfish, slowly but surely Java, etc. I would personally give Sun a good grade.
Open Source can be good for business, huge, large, and small. A bit off topic, but: while I earn most of my living consulting on (unfortunately) closed source projects, I almost always try to initially talk my customers into at least considering Open Sourcing all or parts of development projects. I believe that software development should be done in the least expensive and highest quality way possible: better for almost everyone to drive down the cost of software development; I argue that the less expensive that useful projects are, then more projects get funded. Also, about an hour ago, I received a small grant from someone in Europe to convert one of my LGPL projects from Java to Pascal/Delphi :-)
I give Star Division all the credit for Star Office.
Sure, Sun decided to come along after it was pretty much finished
and but it out and then give it away. However, it was the Germans
that actually built it from the ground up and gave us something
useful. They had the vision and the interest in creating the thing
in the first place. They chose to support us.
It wasn't 'ideal' in terms of free software zealotry but they were
there at the table with us at least showing us enough respect to
acknowledge us.
Giving Sun the main credit for Star Office is like saying that Gates invented the internet.
A Pirate and a Puritan look the same on a balance sheet.
I got Jython working on my new machine, and it's considerably better. Like, almost on par with CPython - modulo the fact that it's still quite behind feature-wise. It might be the JVM (1.5 before vs 1.6 now) helping. Also, IronPython has gotten worse at generators between Alpha6 and Beta2 somehow.
Test machine is 2GHz Core 2 Duo iMac with 4GB SDRAM running MacOS 10.5.3.
awatts@platypus:~$ python2.5 timerseqs.py
2.5.1 (r251:54863, Jan 17 2008, 19:35:17)
[GCC 4.0.1 (Apple Inc. build 5465)]
forStatement => 4.59940099716
listComprehension => 2.38550591469
mapFunction => 1.92037510872
generatorExpression => 3.14438390732
awatts@platypus:/Applications/IronPython-2.0B2$ mono ipy.exe ~/timerseqs.py .NET 2.0.50727.42)
2.5.0 (IronPython 2.0 Beta (2.0.0.2000) on
forStatement => 9.48001098633
listComprehension => 6.89910125732
mapFunction => 5.57528686523
generatorExpression => 19.7973022461
Note: Jython doesn't support generator comprehensions ./jython ~/timerseqs.py
(With Java 1.6.0)
awatts@platypus:~/jython2.2.1$
2.2.1
forStatement => 6.0929999351501465
listComprehension => 3.7769999504089355
mapFunction => 1.6570000648498535
Speaking as a Sun employee: you're welcome.
But do remember that there's an element of self-interest in this open-sourcing strategy. It's all part of our fiendish plot to sell people hardware and services.
Take Solaris, for example. By opening it up, we do lose the income we would have had from selling it to people. But that's been dwindling anyway, as Solaris loses ground against Linux and Windows. By opening up the OS, we make it a better product through user contributions, and encourage its spread. More Solaris users means more people who will seriously consider out products and services.
Of course, even Linux and Windows people should be looking at us anyway, since we are now serious about products that run those OSs. (I work on documenting several of them.) But if you're already a Solaris user, then your options go beyond our x64 systems to the systems that are still the core of our business: the SPARC machines.
There are many reasons SPARC systems have been losing ground. But a big one is that they don't run "standard" operating systems. Promoting Solaris through open-sourcing (and through other means, such as supporting it on other vendor's hardware) drastically changes that particular equation.
I thought that Sun paid a lot of money for StarOffice, and then Sun open sourced it.
Isn't this right?
First of all, the question is not why they didn't use the GPL, but why they didn't use one of the many GPL-compatible licenses.
Second, a license that prevented programs from running on GNU/Linux would (by definition) not be an Open Source license.
Third, I suspect the GPL is the Open Source license *most* court-tested.
Fourth, Linux's GPL license does not prevent any codec from running on it. It's the authors of the codecs and patent holders that do that.
And finally, the GPL hurts Linux's stability? Truly it is a powerful license, but I never imagined that it had such capabilities...
"Not an actor, but he plays one on TV."
In 2000-2001
1. They screwed up by announcing the end of line for Solaris on X86
2. They screwed up by refusing to offer X86 hardware.
3. They screwed up by not offering Linux on any of their hardware
4. They screwer up by not open sourcing Java, Solaris, and other goodies.
In the end, they are trying to correct all those errors, but I wonder whether doing that 7-8 years later means that they missed a golden opportunity to become a leader in the Linux and Unix software and hardware market (including on X86).
Open that code up and consider that a significant part of a complete, sincere apology.
Twitter supports and protects racists - by smearing their critics with the "Hate Speech" label.
For a while, I would agree with the above, but I'm seeing a lot of signs of life out of Sun lately. They really get open source software, and are making money off of it. Simon Phipps this week at Jazoon '08 noted they're making more money off OpenSolaris these past couple years than the past 8 years combined.
It's very easy to pick at a company's decisions -- but it's really hard to turn around a huge company with an entrenched culture; other UNIX players weren't pure plays and are so diversified that it's both easy to hide their own problems (how's HP-UX doing?) or to entice hardware purchases because of broader relationships & bundling (IBM is classic at this).
Sun still has a lot of runway ahead of it. $3B in cash. $13.8b in revenues a year, which is UP $2.8b from 2005. The recent quarter problems are concerning but in context were a 0.5% drop in revenue year-over-year. Yes, it's very bad that they're not very profitable, but let's put it in context -- their big losses were 5 years ago or more. Apple was in much, much worse shape in the mid 90's. Motorola just lost $2.1 billion in revenue from their mobile phone division.
Sun is drifting slowly towards death, but, as they say, the green mile is sometimes quite long.
-Stu