2. MAD, MAD Magazine, and GOM: Don Boettner's and Bruce Arden's recollections

posted Nov 19, 2010, 6:06 PM by Jeff Ogden   [ updated Nov 19, 2010, 6:52 PM ]


On Jul 13, 2010, at 9:37 PM, Donald Boettner wrote:

1. GOM - I have a printed copy of the GOM Manual.  I also think I have it on a CD of saved stuff.  I will look.

2. I think $message was the only thing written in GOM.   When I wrote the command macro processor, I wrote a prototype in SNOBOL (actually SPITBOL) and then did the final version in PLUS.

3.  I just located in my archives a computer output listing dated 06 May 1960 under id R02DN (my course id for Bernie's course) which is a failed compilation and so has the Alfie picture.  (and I have a couple other Alfie ones)   This is a little large for my scanner.  I might try photographing it first and seeing if I can get sufficient detail and contrast that way.

4. Some words on the reason for Alfie and his subsequent disappearance in later years;   Bruce Arden would know better if I have some of the things mixed up:

MAD was written for an IBM 704 computer with 8192 words (36 bit) of main memory and 8192 words of high-speed drum storage.   It was written as a three-pass compiler.  Part I (Bob Graham) did the lexical scanning and generated the triples (DSCAN and CMPMTX for those who remember) and wrote the triples on an intermediate tape and any symbols encountered on the drum.  Part II (Bernie Galler) read in  the symbols from the drum, condensed the table and assigned addresses.  Part III (Bruce Arden) read in the triples from the intermediate tape and using the condensed symbol tape, generated the binary executable.

Most errors were caught in Part I, but there was not enough room for the compiler code and the full text of the error messages.  So Part I was actually in two core loads on the system tape:  the first did the compilation and if it  found an error, it put in a flag and an error number:
   ***************************** ERROR NR.      05172
after the offending source statement.  Then at the end of that phase, if it had detected errors, it read in the next core load which printed a page with Alfie at the top and the text of the error messages:
       5172 INTEGER CONSTANT TOO LARGE.
 (if no errors were detected in Part I, the second core load was skipped and control passed to the next core load which was Part II.)

When the 7090 was installed, it had 32768 words of main memory, and so this indirection was no longer necessary, and MAD was changed so the full text of the error messages was printed following the offending statement.  So there was no reason for the A.E.Neuman picture any more.

I will let you know later about the GOM Manual and the Alfie scan.

     -don


MAD Manuals

There are a number of MAD, MAD/I, and GOM manuals available on the Web, see the Reference section of the Wikipedia article on "MAD, MAD/I, and GOM".


Alfie: A portrait of Alfred E. Neuman as produced by the MAD Compiler in 1960


Don did scan his c. 1960 Alfie printout. Here is the result, which is included in the Wikipedia article on MAD:

   


Bruce Arden remembers a bit about Alfie too:

On Jul 12, 2010, at 11:24 AM, Bruce Arden wrote:

Jeff,

 . . .

I sent the request to MAD magazine for the use of their title, including the title font with the satyrs pursuing the buxom bacchantes. This was used for a while before it was changed due to someone's complaint about propriety. There was also a typo that raised hackles later. I received MAD's approving response and must have it somewhere unless it was a casualty of some move.

Bruce

Comments