Linux advocate: More SCO evidence flawed

But company disputes claim

By Robert McMillan
IDG News Service

August 20, 2003  

The SCO Group is zero for two in its efforts to prove that its Unix software was illegally copied into the Linux operating system, according to Linux advocate Bruce Perens, who on Wednesday said he traced a second example of SCO's disputed code and that it was lawfully included in Linux.

The revelation came the day after Linux enthusiasts claimed to have proved that SCO's first public example of copyright violations in Linux was illegitimate. 

SCO revealed the two snippets of code during a keynote address at its annual SCO Forum user conference in Las Vegas this Monday. The first snippet, which SCO claimed was copied line by line from Unix into Linux, was published by the German publisher, Heise. The second example, which SCO claimed was an "obfuscated," or a nearly identical copy, was obtained from SCO by IDG News Service.

At the keynote, SCO's senior vice president of SCOsource Chris Sontag claimed the snippets were just two examples of many intellectual property violations his team had found in the Linux source. "We've been able to find these little needles in a Mount Everest-sized haystack," he said.

But these first two examples can be traced to the open source BSD (Berkeley Software Distribution) Unix, and not to SCO's AT&T Unix source code, and both are legitimately included in Linux, Perens said.

"I think that these are probably the best examples that SCO has to show and they're awful," said Perens. "They would not stand up for a day in court."

In March, SCO sued IBM claiming that Big Blue had inappropriately contributed code to Linux. Since then, SCO has widened its allegations about the nature of the improper contributions to Linux, and now maintains that source code belonging to SCO has been copied line-by-line into the Linux operating system.

SCO was countersued by IBM in early August, and has also been sued by Red Hat.

"The obfuscated code example is not SCO's property," said Perens. "It was developed by the Lawrence Berkeley Lab in 1993, under funding of the US Government. The code was added to SCO's version of Unix in 1995 or 1996, he maintained. "SCO took (the BSD) source code, lost the attribution, and now believes it's theirs."

SCO disputed Perens' claims. "We're the owners of the Unix (AT&T) System V code, and so we would know what it would look like," he said. "Until it comes to court, it's going to be our word against theirs."

The obfuscated sample, which contains networking software, could have been legitimately copied in the Linux source code, because it has been released under a BSD license, Perens said.

But the code was created from scratch and not copied from any version of Unix, according to the Linux developer who contributed it.

Jay Schulist, a senior software engineer with Pleasanton, California's Bivio Networks says he wrote the 500 lines of code in 1997 as part of a volunteer project for the Stevens Point Area Catholic Schools in Wisconsin. "I used it for helping a local school district in my home town to connect their old Apple Macintosh machines to the Internet," he said.

Schulist wrote the code, based on the publicly available specifications created by Lawrence Berkeley Labs, he said. He has never seen the AT&T source code, he added.

The Linux hacker expressed surprised that his contribution would be singled out by SCO. "I have no idea why they would even chose my code," he said. "If they had done any research at all, they would have realized that there was no other way to implement the actual filtering engine."

Linux creator Linus Torvalds said he had no plans to delete Schulist's code from the Linux kernel. "I'm not removing Jay Schulist's code," he said in an email interview. "We can show where it came from, and there's nothing strange going on there."

Perens' code analysis can be found here: http://www.perens.com/SCO/SCOSlideShow.html

Copyright © 2003