Date: Tue, 10 Dec 1996 15:19:49 GMT Server: NCSA/1.4.2 Content-type: text/html Last-modified: Sat, 28 Sep 1996 03:13:54 GMT Content-length: 6023
Here's a narrative resume, in reverse chronological order.
I'm new to UW, having arrived this summer from Illinois College. IC is a small liberal arts college in historic Jacksonville, Illinois. Founded in 1829, IC was the first college in the state to grant a degree. Perhaps its best-known graduate was William Jennings Bryan. There were other famous visitors in the early days, including Abraham Lincoln (ask me sometime about the Lincoln chair) and Stephen A. Douglas.
At IC I taught practically every computer science course in the catalog.
Before that, I was at Arizona State University in Tempe, AZ, doing a Ph.D. My dissertation was about neural network speech recognition. Languages and linguistics, computational or otherwise, have long been an interest of mine.
Before drifting into ASU's graduate program, I had worked for Honeywell for 11 years, first in Washington D.C. and then in Phoenix.
My job with Honeywell in D.C. was customer support at the international headquarters of a notorious and demanding non-profit organization.
For most of the Honeywell years in Phoenix, I was an operating systems developer. The OS was GCOS-8, a mainframe system supporting multiprocessing, multiprogramming, transaction processing, and large database systems (more about GCOS-8).
Before going to Honeywell, I worked as an applications and systems programmer for the U.S. Maritime Administration, a block from the White House. The District of Columbia was the most interesting place I've ever lived (outside of a summer in Munich) until coming to Seattle (unless you count that year in New York which I never tell people about and which cannot be detected from my resume).
Before D.C., I did a Master's (math) at the University of Kentucky, and a B.S at Kent State University, a few miles from the Ohio town where I grew up. In fact, from the windows of my junior high school classrooms I could see buildings of the KSU campus. Now we're practically back to my idyllic childhood -- a good place to stop. As you might deduce, I'm old enough to remember 3-cent stamps, 48-star flags, the Hungarian uprising, and Sputnik.
Aside from computers and languages, music has played a big part in my life. Even before moving to Seattle I had made four trips to Seattle to see and hear Wagner's epic Der Ring des Nibelungen (plus three trips for other reasons). Since those visits have usually been in July or August, I have developed the firm impression that Seattle has a delightful Mediterranean climate not unlike Nice or San Diego. Nothing has occurred since my arrival July 22 to shake this belief.
More about GCOS-8 (the OS) and
DPS-8 (the hardware architecture). GCOS, its history and internals,
is unfortunately not well known (that's the book I should have
written). It was comparable in many ways to IBM's MVS. Our hardware
architecture, particularly memory management, was more advanced
than that of the IBM 370/3090 line (our main competitor), and
probably influenced the design of the Intel 386 (another untold
story); GCOS-8 had in turn been influenced by Multics, to which
Honeywell owned the commercial rights (a story which has been
told and which has an unhappy ending).
Each native process sees memory as a variable set of unforgeable
"descriptors" which give access to protected virtual
address spaces ranging in size from one byte to many megabytes.
Separate access keys are associated with each descriptor. (This
segmentation scheme is almost completely orthogonal to a virtual
memory arrangement with two levels of page tables.) It was possible
under this scheme to give each global variable and each activation
record its own address space, making constants truly unmodifiable,
giving hardware protection against bounds and pointer errors,
providing a secure method for passing parameters by reference,
and so forth.
There were many other noteworthy innovations in the system, many
of them going all the way to the predecessor non-virtual system
of the late 1960's: GCOS-III, which started life as GECOS. The
GE originally stood for General Electric, which entered the mainframe
business in the 1950's but sold out to Honeywell around 1968.
Among many features that can be traced back to GCOS-III (or earlier?)
are: support for symmetric multiprocessing (up to eight CPUs);
databases (the CODASYL network database model derived from a Honeywell
product); on-line support (time-sharing and transaction systems
which both which ran rings around the analogous IBM products);
a secure and flexible file system, etc. etc. Of course, one can
find blemishes and limitations, too, especially with the benefit
of hindsight.
Shortly after I left Honeywell, the company spun off its computer division into an independent corporation that was owned jointly by Honeywell, NEC, and Bull (France), all of whom had historic ties to or vested interests in various of Honeywell's computer products. Since then, the spin-off company has evolved into a subsidiary of Groupe Bull called Bull HN Information Systems, Inc.