Slashdot Mirror

One might argue that Telebit's spoofing of the UUCP-g protocol (about as inefficient as XMODEM when not spoofed) let the uucp folks get lazy and not bother to implement a real streaming protocol (such as ZMODEM), thus marginalizing uucp to those who could afford that one particular expensive brand of modems

One might, but then one would be wrong. Allow me to introduce you to several other protocols supported by Taylor and several other UUCP suites:

http://www.airs.com/ian/uucp-doc/uucp_7.html

Several sites experimented with "G"; UUNET somewhat supported it.

"x" became quite common in the land of EUNET and a variety of other places.

"t" was used extensively in the early years of the commercial Internet where AUPs were not an issue.

However, "g" spoofing worked very well for the transfer of news batches (and ftpmail transfers), and could beat magtape-via-postal-services to distant places.
Indeed, in the early days of international dial-up SLIP, BSMTP over "g" over Telebit's PEP usually performed best for anything but frequent small emails (frequent enough to build up a uucico queue, but infrequent enough not to be stuffed into a batch).

The key point here is PEP, which itself was not geared towards bidirecitonal traffic such that its handling of overlying streaming protocols was often particularly poor exactly when PEP had the best throughput gains over e.g. V.22bis/V.32/V.32bis (even with V.42).

Eventually the international connectivity programs sponsored by the NSF and a variety of other agencies in the USA (working with counterparts abroad) allowed the migration to "t" over TCP/IP over a variety of synchronous lines (HDLC/PPP) or TCP/IP over X.25. UUCP-over-the-Internet ultimately was replaced by NNTP/SMTP directly over TCP/IP as the news and mail transfer agents began to support pipelining, which became competitive with compressed UUCP batches.

For comparison with Fight-o-net's history, "g" was effectively dead for any substantial site -- pace UUNET, whose customers were paying for it -- by 1991-2, relegated to niche long-haul (and satellite) applications using Telebit PEP + "g" spoofing; "G" did not last much longer, although "t" and "x" were still seing some serious use in 1993-1994.

No, it is fundamentally broken in places. It's not a matter of library support. It's a matter of the compiler taking those libraries and totally fucking with their intended purpose because the language spec lets them and it can improve optimizations.

But I'm glad idiots have opinions. It gives people like this guy (video) something to do, by educating you. What ends up happening is that compilers end up presenting non-standard extensions. I've used an embedded language nesC that would provide atomic blocks and implement them by disabling interrupts (not a great idea when avoidable, but passable if you know the drawbacks).

No, it'll run GNU.

In fact, I have seen the code under NDA.

But I don't need to do that in order to say that it is different code. SCO is saying that already--they are just saying it in a (possibly deliberately) confusing manner.

SCO is claiming that code was copied directly from Unix to Linux. That is the code they are showing under NDA. SCO is also claiming that IBM/Sequent has contributed code to Linux, in violation of IBM/Sequent's contract with AT&T.

I hope that you agree with that.

Now, think about it: the IBM/Sequent code is not in Unix. Why would it be? SCO didn't write it, nor did any previous owner of the Unix source code. IBM certainly didn't give it to SCO. Why would they? For IBM, it's a competitive advantage.

So when SCO claims that code has been copied directly from Unix into Linux, they can not be talking about the IBM code, as that code was never in Unix in the first place.

QED.

I signed the NDA and saw the code. Here is my writeup.

The scenario you describe did not occur for a few reasons.

First, I signed the NDA in good faith. I knew going in that I was not going to be permitted to disclose the code. Those were the ground rules which SCO set, and it was not my intention to cheat them.

Second, SCO is demonstrably a litigious company. Were they to sue me, that would be a major problem for me. Were they to win a lawsuit, I could lose everything I own and have my wages garnished for eternity. While it's true that it would hardly be worth their while to sue me, the level of risk requires cautious behaviour on my part. Basically, I want to be sure that if the code which SCO showed me is removed from the Linux kernel, that there is absolutely no reason to think that I had anything to do with it.

Third, SCO only showed me one example of what they claimed to be direct copying. They claimed that they had many other examples which they were not going to show me. So even if I were to quietly reveal the one example they showed me, it would not affect their claims significantly. Of course, it is possible that they are lying about having other examples. But since SCO's claims in general rely on FUD, removing one instance of potential direct copying, when there are other claimed instances, would not materially lessen the FUD.

I can't really speak to your suggestion that people routinely violate NDAs. I've never knowingly violated one. Aside from any considerations about keeping my promises, if people became aware that I had violated an NDA, I think it would be quite a bit harder for me to find my next job.

I saw the code, and I didn't say that their claims aren't FUD. I said it wasn't clear, and that the one piece of allegedly copied code they showed me proved little. See my writeup at the Linux Journal or on my own server.

BTW, anyone try consoling into a modern SPARC with USB ports, or are they only for peripherals?

I do just this on a quasi-regular basis; the console is a Rev. D iMac with a KeySpan serial thingy and cu(1) from Taylor UUCP, and the SPARC is a headless Sun Ultra 1 running NetBSD.

BTW, anyone try consoling into a modern SPARC with USB ports, or are they only for peripherals?

I do just this on a quasi-regular basis; the console is a Rev. D iMac with a KeySpan serial thingy and cu(1) from Taylor UUCP, and the SPARC is a headless Sun Ultra 1 running NetBSD.

Here is an introduction to automake/autoconf.
--
Whether you think that you can, or that you can't, you are usually right.

Infamous early Cygnus employees include Fred Fish (Amiga & later BeOS free software god), Sean Fagan (general troublemaker on the net), Tom Jennings (author of the original FidoNet software), Brendan Kehoe (G++ maintainer, author of Zen and the Art of the Internet, one of the first open source books and one of the first popular books about the 'net), Steve Chamberlain (extraordinary speed hacker, creator of Cygwin which built on DJGPP - he's now at TranceMeta), Ian Taylor (author of GNU/Taylor UUCP), and others too infamous to mention.