Slashdot Mirror
Usenix President - Linux Needs Better Paper Trail
Anonymous Coward writes "Usenix Association president Marshall Kirk McKusick is a veteran of BSD's intellectual property scuffle with AT&T in the 1990s, and he's got some thoughts and advice for the keepers of the Linux kernel going forward, commenting: 'There isn't a well-documented ownership trail with Linux. So, they have opened themselves up to a swamp of 'he said-she said' about where code came from'."
Also, I would imagine that pretty much every kernel code submission is traceable to it's submitter. As far as I know, every specific line of code that has been brought up by SCO has been tracked down and attributed to it's submitter. Beyond that, there's really no way for BSD, Linux or anybody else to _know_ that the person submitting a patch really owns the copyright to it, or is acting as an authorized agent of their employer who owns the copyright to it. At some point, there is good faith. Yes, a well-documented paper trail would be nice, but requiring patch submitters to submit signed documentation with their patches would place an immense administrative burden on somebody, and it wouldn't prove that no copyright infringement has occurred, it would just blame-shift to whoever submitted the patch. I don't think that would legally remove the possibility that an unscrupulous company could go fishing for damages, a la SCO. It would also effectively remove the bazaar-like openness that Linux has, in contrast with more closed, insular projects with their rigid committer lists and uberpolitical machinations (XFree86 anyone?).
But I guess from a PR perspective this guy has a good point. Having some big pile of papers to point to and say "look, this documents that all contributers have copyright to their patches, and every line of code is accounted for" - this might help dissuade outrageous claims like SCOs and allay the fears of the business community, which likes to know that there are reams of bureaucratic documentation proving that the code is clean.
And maybe *you* should read RTFA (the McKusick interview)? He says explicitly that the paper trail was lacking "until recently" (by which he means the switch to BitKeeper). Maybe you should also learn some respect for people like McKusick who've been hacking free Unix since back when Linus was a kid. Among other things, this guy pretty much invented the modern Unix "fast file system", from which ext2 takes a lot of ideas. More recently, he's been responsible for softupdates in UFS (gaining the speed benefits of async mounts without compromising filesystem integrity in case of crashes).
How many closed source companies copy code from various places? I would say open source is the least likely place people will do this considering how easy it is to get caught.
Anyone out there have personal knowledge?
We all know Linux hasn't been in any sort of a version control system since version 2.2 after which the issues started alledgedly creeping up.
Marxist evolution is just N generations away!
It's very easy to document where code did come from. But it's virtually impossible (if not 100% impossible!) to document that code did not come from any commercial source. By definition, to "prove" that any given piece of code didn't come from a commercial source, you'd have to take every single piece of commercial source code written up to and including the day of the disputed source's release, and grep it.
Honey, I shrunk the Cygwin
No matter how convoluted the system you propose to "track" this stuff, it will *always* come down to whether you beleive or can trust "the first order contributer".
If we knew where every last keystroke came from, there would still be the "bob is lying, that keystroke didn't come from him, he stole it from his bos/frind/company/disassembly-fo-windows or whatever. Or worse, he typed in the code but he got the idea from watching the wonderful-world-of-Disny while reading Cryptonomicon so Eisner and Stephensen are the inventors and deserve X in consideration.
Many jobs worth doing are only worth doing to a certian standard of completeness. The problem with the porely-named Intellectual Property domain is that, reguardless of whether ideas want to be free or $40 a barrel, the boundary and origin of all ideas is undocumented-bastardary at best.
All works of any creative mind are, at least in part, stolen from the fertile field of experience.
There is no fixing that, and the supposition that all the progenitors of what came before do *NOT*, a-prioria, deserve recognition and a stake.
Turning the provenance of each line of code into a preverse kind of Oscal(tm) acceptance speach *still* wont insure that someone isn't slighted somewhere.
"I'd like to thank the academy, and my third grade comp-sci teacher for this for-loop, without them I would have never understood that pre-increment saves a temporary. And of course a shout-out to the CPU manufacturer, without whom I'd have never had a chance at direct increment of non-register memory. And of course my Mom, who never let me leave the table without eating all my peas; if it weren't for her I'd have never learned the value of bounds-checking in the completion of a problem domian. I know I'm forgetting someone, but you have all been so wonderful..." -- Rob White, Linux Kernel 6.2 Changelog for kernel.c line 722.
NOTE: The Above attribution is Under Dispute from the GCC board of optimizers for failure to credit the optimizing community's efforts in envisioning the need for loop unroling and the value of peep-hole allocation of registers...
Really, how bad does "intellectual property" have to get before people get it into their heads that the Founding Fathers *DID* understand that you cannot own an idea. The absence of computer science from their accumen doesn't mean that these topics are sacrocant, wholly new, and innumerable to that prior understanding.
Clue please people...
Innocent people shouldn't be forced to pay for inferior software development.
--"Code Complete" Microsoft Press
pencil and paper: n.
An archaic information storage and transmission device that works by depositing smears of graphite on bleached wood pulp. More recent developments in paper-based technology include improved 'write-once' update devices which use tiny rolling heads similar to mouse balls to deposit colored pigment. All these devices require an operator skilled at so-called 'handwriting' technique. These technologies are ubiquitous outside hackerdom, but nearly forgotten inside it. Most hackers had terrible handwriting to begin with, and years of keyboarding tend to have encouraged it to degrade further. Perhaps for this reason, hackers deprecate pencil-and-paper technology and often resist using it in any but the most trivial contexts.
Courtesy of the Jargon File. See, I know what paper is!
If my answers frighten you, stop asking scary questions.