Tuesday 3/31/98 7:39 AM John Gilmore John Young J Orlin Grabbe Reading the stuff on the pals at http://www.jya.com/pal-nukes.htm http://www.research.att.com/~smb/nsam-160/pal.html jogged my memory. After we completed the Forth implementation of the Missile Secure Cryptographic Unit at least Hal Pruett, Duane DeWerff, myself, and perhaps several others were invited to NSA at Fort Meade in about 1985-6. NSA employee Bill Legato assailed me for using Forth for the implementation of the MSCU. Legato hated Forth and relayed his hate to Sandia management. This is one reason I was replaced as project leader by Tom Evans who was later replaced by Eric Disch. Possibly surprising, however, is that NSA sent Forth expert Don Simard to visit Sandia for several years. Kent Parsons commented to me that he thought Simard was an NSA spy sent to see what we were up to at Sandia. Simard, along with Bill Goldrick, were the two reviewers of my SAND report now seen at http://jya.com/whpfiles.htm EE Times, March 30, 1998 page 8 A Java chip being to ship in quantity SAN FRANCISCO - Patriot Scientific Corp. surged to the head of the silicon pack at the JavaOne conference last week, showing the PSC1000 RISC processor, which is shipping in quantity. Well be selling the chip for under $10 in volume and aiming at deeply embedded, low-power applications such as handheld devices and factory-floor control systems, said Phil Morettini, vice president of Patriot (San Diego). Competing offering from LG Semicon and NEC Corp., each based on Sun Microsystems Inc.s picoJava core, are not yet available, although select customers are said to have samples. Separately, Rockwell Collins Inc. (Cedar Rapids, Iowa) said that it has decided not to sell its Java-specific CPU on the merchant market. The company unveiled the innovative processor last fall. Here is their web site. http://www.ptsc.com/ Charles Moore, inventor of Forth, is author of the Shboom Java native code compiler for the PCS100. Rockwell built a Forth-like chip many years ago. I met Randy Dumse, who then worked for Rockwell, at a trade show. I got a copy of the Rockwell R65F11 chip manual from Dumse. Dumse quit Rockwell to start NewMicros seen at http://www.newmicros.com/ Click on history to see http://www.newmicros.com/history.html Of course, the Harris RTX 2000 chip used by NASA is a descendant of Charles Moores Novix Forth chip. http://groucho.gsfc.nasa.gov/forth/ Ballard, I was told by Jeff Allsup of Wood Hole, used polyForth to discover the wrecks of the Titanic and Bismarck. A central debate is whether a high-level language should be used for a weapons programmer. The implementer must be able to account for each bit. In Forth I can. Possibility exists that an disgruntled implementer might 'spike' a weapon so that it went off at an unintended time. Sandia supervisor Don Schroeder was an OUTSPOKEN critic of the T 1563 nuclear bomb controller. SCHROEDER,DONALD H. (505)845-8409 DHSCHRO (505)844-9478 0519 DHSCHRO@sandia.gov The T 1563 did not work very well. One reason is that it was written in about 112,000 lines of assembler. NSA was reluctant to approve use of an operating system or even a high-level language for crypto implementation for the reasons that security could not be guaranteed. Sandia did not have the resources required to write its own secure operating system and high-level language. This is why we used the Forth operating system and high-level language. Schroeder division employee Mike Sharp SHARP,MICHAEL W. (505)845-8444 MWSHARP (505)284-3850 1138 MWSHARP@sandia.gov wrote a T 1563 implementation in, I think C or Pascal. Henry Newgass modified the transient part of command.com so that on power-up a PC directly entered Sharps program. Jerry Allen, who also saw the merits of using an operating system and high-level language, funded Neugass. ALLEN,DOUGLAS J. (505)845-9624 DJALLEN (505)844-7593 1201 DJALLEN@sandia.gov I wrote the purchase order. I learned several months ago that Sandia built a PC version of the T 1563 called the T 1663, I was told, which was supplied to the Air Force. The threaded code - compiled native code battle continues. I am sort-of happy to be back [somewhat] doing MASM code - along with a bit of Forth and Forth assembler on the 80c32. I see that J Orlin Grabbe is linking Permissive Action Links at http://www.aci.net/kalliste/ I read Both strong links must be closed electrically -- one by specific operator-coded input and one by environmental input corresponding to an appropriate flight trajectory -- in order for the weapon to be armed. George Dullek and Jon Bryan worked together on accellerometers for flight trajectory sensing. Dullek and Bryan are Forth programmers. DULLECK JR.,GEORGE R. (505)844-2628 GRDULLE (505)844-8480 1073 GRDULLE@sandia.gov BRYAN,JON R. (505)844-2015 JRBRYAN (505)844-6161 1003 JRBRYAN@sandia.gov Sandia had a division devoted to crystal oscillators. The oscillators for the micro occasionally fail to start - a bad scene if this happens when trying to arm a nuke. Dick Adams, also in department 2300, was in charge of oscillators. Adams never seemed to many of us too interested in his job. Adams took an early voluntary separation package from Sandia with a large educational retraining incentive. And went to cooking school in San Francisco, Bob Wayland told me. The real world again. Thanks to good research, some whistleblowing, and Internet details of fuzing the US nuclear arsenal are being revealed. Lets ALL hope for SETTLEMENT of this unfortunate mess before it gets WORSE. I want my money and I am out of this mess. Later bill