To read comments to this article, go here
|SCO Says SGI Isn't Doing Enough, but What Would Be? --And Details on SGI's Code Comparison
Sunday, October 05 2003 @ 08:24 PM EDT
It appears nothing SGI has done is enough to satisfy SCO, and in fact Darl McBride's statements make one wonder if anything short of removing XFS would be enough, and because that isn't going to happen, it looks like the two sides are going to legally rumble:
The SCO Group has insisted that changes made by Silicon Graphics (SGI) to some of its Unix code will not be enough to prevent termination of SGI's Unix licence.
SCO plans to revoke SGI's Unix licence even though the latter claims to have removed all potentially offending code from its XFS journalling file system, now in Linux. But this does not go far enough, SCO has told vnunet.com .
The licence is due to be terminated on 14 October, two months after a warning letter was sent to SGI complaining that it had allowed code to be transferred from SCO's Unix System V into Linux.
The letter, dated 13 August, claimed that SGI subjected SCO's source code to "unrestricted disclosure, unauthorised transfer and disposition, and unauthorised use and copying". . .
"We don't believe that SGI has taken all of the steps necessary to cure all of the breaches, and in fact in our letter to them, we state 'SGI's breaches of these agreements cannot be cured'."
Laura DiDio has said publicly on October 2, 2003, that SCO has been aware of "blatant SGI violations" for a year. That raises a number of issues. First, if they knew about the "violations" since October of 2002, why didn't they do anything about it until now? And, more, since SCO claims that they set up their stock plan in advance of any knowledge of legal action, that would appear now not to be the case, if what DiDidio says it true.
As Groklaw noted back in August, second story down on the page:
Bloomberg News has an article, appearing in The Salt Lake Tribune, reporting that Darl McBride says that SCO's CFO submitted a sales plan in January "months before legal action was contemplated", presumably as proof that there is no connection between the stock sales and the lawsuit:
Chief Financial Officer Robert Bench began the selling by SCO insiders, four days after SCO filed the suit against IBM. Bench is selling to help pay a $150,000 tax bill, McBride said. Under the Sarbanes-Oxley law, companies are no longer able to loan executives money to pay taxes or other expenses. [emphasis added]
Bench submitted a sale plan in January, months before any legal action against IBM was contemplated, McBride said. His agreement called for the sales to begin on March 8. He planned to sell 5,000 shares a month for the next 12 months, according to the plan."
But if they knew about the "violations" back in October of 2002, and hired Boies in January, it does raise some questions in my mind about the January plan and just how preset it actually was. I hope the SEC reads Groklaw, because they are in a better position than I am to know what this all adds up to and can do something about it, if it feels it would be appropriate.
Another thing: Was SGI's contribution of XFS to Linux a secret they only now just discovered? Hardly. Aside from Linux code being publicly viewable at all times by SCO, and the fact that they have distributed their own distribution of Linux themselves -- with XFS, by the way, which means they made money from it and knew it was there, as you can see for yourself in this SCO product brochure which mentions XFS by name, as well as NUMA and JFS -- consistently throughout the time period in question, SGI first announced they would be donating XFS to Linux back in May of 1999:
InfoWorld: SGI extends an OS crown jewel to Linux effort
May 21, 1999, 12 :03 UTC
SGI on Thursday here at Linux Expo announced it will make the heart of its Irix operating system -- its 64-bit journaled XFS file system -- available to the open-source community.
"XFS will help Linux get past some of the limitations it has today, like scaling and the 2GB file limit. Our goal is to make Linux as capable as [SGI's] Irix in terms of scaling and number of processors it can handle," said Harris Shiffman, strategic technologist for SGI in Mountain View, Calif.
"XFS will enable Linux to scale high enough to handle file systems as large as 18 million terabytes of data and files up to 9 million terabytes, SGI officials said."
It wasn't until about a year later that SGI legal authorized its release, so it wasn't the case that no one knew what was happening and someone was sneaking code into Linux. It was cleared by SGI legal only after an extensive in-house "encumbrance review and cleanup" process, and if that isn't enough, folks, to clear Linus and open source of any problems with code contributions, exactly what is enough?:
Re: Source code for Linux XFS now available! If you follow this thread, by the way, you'll see the open source process at work. Here's the press release SGI put out at the time:
To : "David S. Miller"
Subject : Re: Source code for Linux XFS now available!
From : Russell Cattelan
Date : Thu, 30 Mar 2000 22:19:58 -0600
Organization : Moo Solutions
> "David S. Miller" wrote:
> > One would think that with a year or so of internal work this sort of
> > stuff would have been cleaned up already. If you guys had released
> > this a year ago things would be much further along than it is right
> > now, I think that keeping it internal for so long was the biggest
> > mistake SGI made about XFS on Linux. Even if it didn't build nor
> > work, releasing 'a source tree' a year ago would have had the whole
> > world contributing to fix it all up and we'd have a working and clean
> > XFS in the tree already.
> > But that's SGI politics for you.
> No, it isn't.
> In fairness to all of us, inside and outside SGI, one must come to
> terms with the fact that we could not publish XFS code before the
> encumbrance review and cleanup were complete. All of us on the XFS
> team at SGI would have liked nothing better than to publish XFS as
> soon as we made the announcement last year.
The green light from legal literally wasn't given till late Tuesday.
I was up till 3 am finalizing the cvs transfer procedure.
The source code release was something we all wanted.
Have to start bracing for all the patches. :-)
> Dan Koren
SGI Empowers Linux With the Availability of XFS 1.0 and Linux Failsafe Plug-Ins SGI Empowers Linux With the Availability of XFS 1.0 and Linux Failsafe Plug-Ins The date on this press release, which isn't on Yahoo! any more but is still on Wayback, is May of 2001, as you can also see from this posting:
MOUNTAIN VIEW, Calif., May 1 /PRNewswire/ -- SGI (NYSE: SGI -news ) today announced the production release of XFS(TM) 1.0 for Linux®, the high-end SGI journaled filesystem software, and code availability of Linux FailSafe(TM) clustering plug-ins. As a leading supporter of Linux and the Open Source community, SGI is once again providing business-critical infrastructure for Linux by porting scalable and robust technology such as XFS 1.0 and Linux FailSafe.
XFS 1.0 for Linux, a journaled filesystem that improves performance and speeds recovery in the event of a system failure, and the code for Linux FailSafe, a commonly used clustering tool to run critical applications, is now available for download on SGI's open source Web site at http://oss.sgi.com/projects/xfs/ and http://oss.sgi.com/projects/failsafe/ . There is no cost for either item.
Linux FailSafe is a high-availability clustering tool that helps companies maintain system uptime and ensure continual access to software and services, while eliminating the risk of system failures on critical applications. Linux FailSafe is architected to scale up to 16 nodes in a cluster with the cluster members sharing storage, allowing multiple servers to assume control of data in the event of a failure. Now with SGI's open sourced Linux FailSafe plug-ins, NFS(TM), Samba(TM) and Apache(TM) applications will be able to failover on system failure and users will still have access to their data.
XFS Journaled Filesystem
XFS, the widely recognized and industry-leading high-performance filesystem, provides reliability and rapid crash recovery without hampering performance. Existing filesystems entail significant limitations in scalability and stability, whereas SGI XFS is the first journaled filesystem that has over a six-year solid track record in production environments. Its mature technology has been proven on thousands of IRIX® OS-based systems as the default filesystem for all SGI customers.
XFS 1.0 for Linux works with the Linux 2.4 kernel and offers unique advantages that other filesystems do not, including:
-- Crash recovery -- fast recovery, regardless of the number of files
being managedLicense and Availability
-- Scalability -- meets the most demanding storage capacity and I/O
-- Space allocation -- enables systems to efficiently scale to support
large numbers of files and directories
-- Complete tool set to support filesystem features
-- Excellent integration with other Linux subsystems
Both XFS 1.0 and Linux FailSafe are licenses under GPL and are currently available at no cost.
SGI provides a broad range of high-performance computing and advanced graphics solutions that enable customers to understand and conquer their toughest computing problems. A strong advocate for open technology, SGI is committed to contributing technology from its core competencies in visualization and high-performance computing to the Open Source community to help solve complex data problems. Headquartered in Mountain View, Calif., with offices worldwide, company information is located on the Web at www.sgi.com .
SGI's xfs So removing code snippets isn't enough, as Blake Stowell says:
To : debian
Subject : SGI's xfs
From : Matthias Berse
Date : Wed, 2 May 2001 11:10:22 +0200. . .
Are there any plans in supporting the usage of SGI's xfs filesystem in
debian? Are there kernel patches available and/or userspace tools
Since yesterday xfs 1.0 for linux is out and under gpl
+-------------created at Wed May 2 11:06:17 CEST 2001-----------------+
In response, SCO's director of public relations, Blake Stowell, told vnunet.com: "Making minor amendments to its XFS file system doesn't cure the breach. SGI must do more as outlined [in the August letter] to cure all of their breaches."SGI Developer Central Open Source/Linux XFS is here, if you want to do your own research. SGI press releases for 2000 onward are here. And
here is a Google SGI XFS search results page.
So, all in all, it certainly appears that SGI has the opportunity to bring its own declaratory judgment action, if it so chooses, thanks to the incredible public statements of McBride and his Merry Men. They keep bringing in new lawyers, but the essential legal problem they have is not, apparently, curable, which is that they talk too much and do too little. They are now faced with explaining to the judge(s) why they never bothered to do anything about XFS since 1999, why removing any conceivably related code snippets that are apparently in the public domain anyway isn't enough, why they distributed Linux with XFS, and exactly why they shouldn't be the defendant in a declaratory judgment action, now that they have made clear that there is an actual controversy, at least from all I see and can understand from the public record. I hope SGI makes the letter public and the contract. Also, I hope SGI brings a declaratory judgment action, and I hope this material is useful to them.
The big news is that SGI has looked at and compared System V code and Linux, and they say they have only found trivial snips that could arguably overlap and even Gartner says this news is not helpful to SCO:
SGI declined to reveal any details on the additional code segments it found, but the fact that its analysis appears to reveal no extensive overlap between the code in Linux and System V is good news for Linux users, according to Gartner analyst George Weiss. They checked it twice, first with Eric Raymond's Comparator and some of their own tools, and then they checked again more thoroughly later in September:
SGI's code comparison was done during September using the Comparator software created by open source advocate Eric Raymond, as well as some other internally developed tools, according to SGI. It compared source code from the Unix System V release 4.1 software that SGI has licensed from SCO with a version of the Linux kernel released this June, SGI said. Looks like SCO's copyright claims just kind of went poof. Now if every other UNIX vendor that has made contributions to Linux does the same thing SGI has done -- and I do include HP in this, if they want to help out -- then that should be that. If anyone thinks IBM hasn't done it already, I'd say they don't know IBM very well. This leaves only SCO's very unusual derivative code theories and perhaps some "obfuscated" code claims, which will likely end up on the cutting room floor too, just like the SCOForum slide show code and now the copyright claims appear to have done. Although there is this very funny quotation in the Vnunet article by SCO's Blake Stowell, in the I'll-huff-and-I'll-puff department:
"Our review was focused on the code we contributed to Linux; however, we did run the Comparator code on the Linux 2.4.21 kernel. The process involves using subjective judgment to review similarities identified by the tool," said Greg Estes, SGI's vice president of corporate marketing, in an e-mail response to questions. . . .
Then in September SGI carried out its more comprehensive comparison. . . .
This comparison revealed a few examples of line-by-line copying, but did not determine whether the code was owned by SCO or in the public domain, according to the letter. "SGI has discovered a few additional code segments ... that may arguably be related to the Unix code," Altmaier wrote. He added that these segments were "trivial in amount."
The fact that SGI has replaced the three code fragments in question does not satisfy SCO, according to Blake Stowell, a SCO spokesman. "These releases have already taken place in Linux," he said. "You still have all these machines out there that haven't applied patches that are still benefitting from this Unix System V code." Well.
Priceless, isn't it?
Thank you very much, SGI. Thank you, Eric Raymond. And to SCO: shame on you. Shame. And may all your public statements come back to haunt you now.
Now, about those invoices for copyright infringement you've been threatening us with...
UPDATE: Groklaw readers have responded with some more evidence and here it is:
http://www.caldera.com/support/docs/scolinux/ipf/relnotes-ipf.html -- Note point 12
More links that may indicate SCOG had knowledge of the origins of XFS and JFS and yet released them themselves:
"This is terrific news and we're happy to have SGI as part of the open source family," said Ransom Love, president and CEO of Caldera Systems, Inc. "There's a great need in Linux for business to have that enterprise-class file technology and storage capability. Daily, we have enterprise customers asking for these solutions - particularly where graphics are concerned. With SGI's contribution and expertise in journaling, throughput and data integrity, we can meet the file sharing/storage needs of those customers with the best technology available. That SGI would make this contribution to the open source community says a lot about their vision and business acumen." - Ransom Love, CEO Caldera
You can read the full release here (dated May 20, 1999).