Author: Robert Hyatt
Date: 20:27:00 03/28/04
Go up one level in this thread
On March 28, 2004 at 13:09:30, Steven Edwards wrote: >On March 27, 2004 at 16:54:54, Robert Hyatt wrote: >>On March 27, 2004 at 11:36:06, Steven Edwards wrote: >>>On March 27, 2004 at 10:24:13, Robert Hyatt wrote: >>>>On March 27, 2004 at 00:17:56, Steven Edwards wrote: > >>>From my CDC programming days, my recollection is that external symbols were >>>limited to seven characters in length, so it's not clear how the BSB symbol was >>>represented in CDC Compass. Maybe it was not an external? Local identifiers >>>had a longer limit (like forty characters)? I never saw the Chess 4.x Compass >>>source, only the Fortran for Chess 3.x. > >>They were truncated. The only trick was to be unique in the first N chars, but >>I don;t remember wheter N was 6 or 8. > >I managed to find a compiler I wrote 25+ years ago in Fortran plus Compass that >produced Compass output. CDC Fortran allowed seven character identifiers (one >more than the standard) and Compass supported this. All local file names were >limited to seven characters, although permanent (cataloged) filenames could have >up to forty characters. > >The primary reason for the seven character limit was the CDC custom of using its >sixty bit CPU word for storing a name in the upper forty-two bits (remember, it >used a six bit display code) and a pointer in the bottom eighteen bits. Seymour >Cray designed some exceptional computers, and their floating point throughput >was remarkable for the time. But the way those CDC machines handled character >data, along with integer multiplication and division, was a joke. > >>I had the chess 4.7 source _years_ ago as >>Harry and I used it for sparring on a Cyber 176 Livermore Labs had... > >I've never seen the 4.x source. But I recall that they used their own set of >Compass macros for nearly the whole project, and I'd probably have done the same >as that assembler was fairly powerful. Indeed, one instructor at the school >wrote a macro package that transformed Compass into an IBM 370 environment >assembler and the resulting binary output was ported to a 370 clone with no >problems. > >Didn't Chess 4.x use only English Descriptive Notation for move I/O? And that >it was the last major program to do so? Yes. But so did we for years, until it became obvious that algebraic was going to "take over". And yes again to the macro question. It made parts of the code easier to read..
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.