Nerd relic: found this in the basement...

[Manedwolf]

My old TRS-80 Model I sits on the storage rack behind where I am now. Complete with monitor, expansion interface, two 180K floppy drives, cassette recorder, and the special carrying cases that were available as an option way back when.
It all still worked the last time I tested everything, but that was ten or so years ago.

There really are a bunch of nerds on this site.
Or maybe nerd relics...

What model of those was big with a built-in monitor and was sort of cream colored? It said TRS-80 on it. We had them in kindergarten, and I remember they weren't that good, very blocky graphics and just stuff like math quizzes and all. Every kid wanted to use the fewer shiny new Apple IIe's instead. But maybe because those had the educational games on them. :)

[RocketMan]

Hmmm....used to sell the silly things back in the day...was it a Model III with a different colored case?  That one found its way into a few schools.
The Model II was a business machine, also with a built-in monitor, but considerably more expensive.  It could even be had with a 5MB or 10MB hard drive if memory serves.  I don't remember what colors it was available in.
I used to have a bunch of catalogs with all that old RS gear in them.  They were pitched just last year.

[Nitrogen]

MY favorite old timey computer story EVER was The Story of Mel, of which I shall replicate in its glory.

         Real Programmers write in FORTRAN.

     Maybe they do now,
     in this decadent era of
     Lite beer, hand calculators, and "user-friendly" software
     but back in the Good Old Days,
     when the term "software" sounded funny
     and Real Computers were made out of drums and vacuum tubes,
     Real Programmers wrote in machine code.
     Not FORTRAN.  Not RATFOR.  Not, even, assembly language.
     Machine Code.
     Raw, unadorned, inscrutable hexadecimal numbers.
     Directly.

     Lest a whole new generation of programmers
     grow up in ignorance of this glorious past,
     I feel duty-bound to describe,
     as best I can through the generation gap,
     how a Real Programmer wrote code.
     I'll call him Mel,
     because that was his name.

     I first met Mel when I went to work for Royal McBee Computer Corp.,
     a now-defunct subsidiary of the typewriter company.
     The firm manufactured the LGP-30,
     a small, cheap (by the standards of the day)
     drum-memory computer,
     and had just started to manufacture
     the RPC-4000, a much-improved,
     bigger, better, faster --- drum-memory computer.
     Cores cost too much,
     and weren't here to stay, anyway.
     (That's why you haven't heard of the company,
     or the computer.)

     I had been hired to write a FORTRAN compiler
     for this new marvel and Mel was my guide to its wonders.
     Mel didn't approve of compilers.

     "If a program can't rewrite its own code",
     he asked, "what good is it?"

     Mel had written,
     in hexadecimal,
     the most popular computer program the company owned.
     It ran on the LGP-30
     and played blackjack with potential customers
     at computer shows.
     Its effect was always dramatic.
     The LGP-30 booth was packed at every show,
     and the IBM salesmen stood around
     talking to each other.
     Whether or not this actually sold computers
     was a question we never discussed.

     Mel's job was to re-write
     the blackjack program for the RPC-4000.
     (Port?  What does that mean?)
     The new computer had a one-plus-one
     addressing scheme,
     in which each machine instruction,
     in addition to the operation code
     and the address of the needed operand,
     had a second address that indicated where, on the revolving drum,
     the next instruction was located.

     In modern parlance,
     every single instruction was followed by a GO TO!
     Put *that* in Pascal's pipe and smoke it.

     Mel loved the RPC-4000
     because he could optimize his code:
     that is, locate instructions on the drum
     so that just as one finished its job,
     the next would be just arriving at the "read head"
     and available for immediate execution.
     There was a program to do that job,
     an "optimizing assembler",
     but Mel refused to use it.

     "You never know where it's going to put things",
     he explained, "so you'd have to use separate constants".

     It was a long time before I understood that remark.
     Since Mel knew the numerical value
     of every operation code,
     and assigned his own drum addresses,
     every instruction he wrote could also be considered
     a numerical constant.
     He could pick up an earlier "add" instruction, say,
     and multiply by it,
     if it had the right numeric value.
     His code was not easy for someone else to modify.

     I compared Mel's hand-optimized programs
     with the same code massaged by the optimizing assembler program,
     and Mel's always ran faster.
     That was because the "top-down" method of program design
     hadn't been invented yet,
     and Mel wouldn't have used it anyway.
     He wrote the innermost parts of his program loops first,
     so they would get first choice
     of the optimum address locations on the drum.
     The optimizing assembler wasn't smart enough to do it that way.

     Mel never wrote time-delay loops, either,
     even when the balky Flexowriter
     required a delay between output characters to work right.
     He just located instructions on the drum
     so each successive one was just *past* the read head
     when it was needed;
     the drum had to execute another complete revolution
     to find the next instruction.
     He coined an unforgettable term for this procedure.
     Although "optimum" is an absolute term,
     like "unique", it became common verbal practice
     to make it relative:
     "not quite optimum" or "less optimum"
     or "not very optimum".
     Mel called the maximum time-delay locations
     the "most pessimum".

     After he finished the blackjack program
     and got it to run
     ("Even the initializer is optimized",
     he said proudly),
     he got a Change Request from the sales department.
     The program used an elegant (optimized)
     random number generator
     to shuffle the "cards" and deal from the "deck",
     and some of the salesmen felt it was too fair,
     since sometimes the customers lost.
     They wanted Mel to modify the program
     so, at the setting of a sense switch on the console,
     they could change the odds and let the customer win.

     Mel balked.
     He felt this was patently dishonest,
     which it was,
     and that it impinged on his personal integrity as a programmer,
     which it did,
     so he refused to do it.
     The Head Salesman talked to Mel,
     as did the Big Boss and, at the boss's urging,
     a few Fellow Programmers.
     Mel finally gave in and wrote the code,
     but he got the test backwards,
     and, when the sense switch was turned on,
     the program would cheat, winning every time.
     Mel was delighted with this,
     claiming his subconscious was uncontrollably ethical,
     and adamantly refused to fix it.

     After Mel had left the company for greener pa$ture$,
     the Big Boss asked me to look at the code
     and see if I could find the test and reverse it.
     Somewhat reluctantly, I agreed to look.
     Tracking Mel's code was a real adventure.

     I have often felt that programming is an art form,
     whose real value can only be appreciated
     by another versed in the same arcane art;
     there are lovely gems and brilliant coups
     hidden from human view and admiration, sometimes forever,
     by the very nature of the process.
     You can learn a lot about an individual
     just by reading through his code,
     even in hexadecimal.
     Mel was, I think, an unsung genius.

     Perhaps my greatest shock came
     when I found an innocent loop that had no test in it.
     No test.  *None*.
     Common sense said it had to be a closed loop,
     where the program would circle, forever, endlessly.
     Program control passed right through it, however,
     and safely out the other side.
     It took me two weeks to figure it out.

     The RPC-4000 computer had a really modern facility
     called an index register.
     It allowed the programmer to write a program loop
     that used an indexed instruction inside;
     each time through,
     the number in the index register
     was added to the address of that instruction,
     so it would refer
     to the next datum in a series.
     He had only to increment the index register
     each time through.
     Mel never used it.

     Instead, he would pull the instruction into a machine register,
     add one to its address,
     and store it back.
     He would then execute the modified instruction
     right from the register.
     The loop was written so this additional execution time
     was taken into account ---
     just as this instruction finished,
     the next one was right under the drum's read head,
     ready to go.
     But the loop had no test in it.

     The vital clue came when I noticed
     the index register bit,
     the bit that lay between the address
     and the operation code in the instruction word,
     was turned on ---
     yet Mel never used the index register,
     leaving it zero all the time.
     When the light went on it nearly blinded me.

     He had located the data he was working on
     near the top of memory ---
     the largest locations the instructions could address ---
     so, after the last datum was handled,
     incrementing the instruction address
     would make it overflow.
     The carry would add one to the
     operation code, changing it to the next one in the instruction set:
     a jump instruction.
     Sure enough, the next program instruction was
     in address location zero,
     and the program went happily on its way.

     I haven't kept in touch with Mel,
     so I don't know if he ever gave in to the flood of
     change that has washed over programming techniques
     since those long-gone days.
     I like to think he didn't.
     In any event,
     I was impressed enough that I quit looking for the
     offending test,
     telling the Big Boss I couldn't find it.
     He didn't seem surprised.

     When I left the company,
     the blackjack program would still cheat
     if you turned on the right sense switch,
     and I think that's how it should be.
     I didn't feel comfortable
     hacking up the code of a Real Programmer.

[lee n. field]

What model of those was big with a built-in monitor and was sort of cream colored? It said TRS-80 on it. We had them in kindergarten, and I remember they weren't that good, very blocky graphics and just stuff like math quizzes and all.

That would be the Model 4, with 80x24 screen (instead of 64x16 like the model iii), 64kb of ram plus a 64kb ramdisk.  Could run TRSDOS or CP/M!

My Model III has been sitting in my Dad's chickenhouse for about 20 years.

[RocketMan]

I forgot about the Model IV.  It's hell losing my...uh, what's that shiny thing?

[GNLaFrance]

OMG, I'm having flashbacks... wait, no, I'm okay, I can't remember the switch settings to boot a PDP-11/44 any more... I guess that's a Good Thing... but I still remember my first Univac fire control 'puter, and how cool it was to get a full load of 64Kb of core in our UYK-20s...

Wish I still had my TI-99/4A and the Expansion Box and all the goodies for it, though.

[ArfinGreebly]

"Burn The Disk Packs"

 

[Cliff47]

I can relate to you, Hutch.  I can remember re-creating the carriage control tapes that got shredded by the central pin-feeds in the middle of the night, swapping control panels on unit-record (punch card equipment for the not-so-old) machines, multi-tasking on two card sorters (083 and 084, with different sort sequences on each), 402/403 tabulating accounting machines, and the occasional stint on a card punch to recreate a deck of header cards for a report.

It's a bummer being old.   Now I did it again, stopped to think and forgot to restart.

Does anybody remember the FUBAR3 pessimizing compiler??

[ctdonath]

Oh, the great story of Mel! That, and a few others ("The Breakfast Food Cooker", "Goto Considered Harmful", "The 'More Magic' Switch", etc.) should be consolidated into a Great Tales of Programming anthology.

Just yesterday I was reminded that from time to time "self modifying code" lives on in great works (such as the core rendering loop in Doom).

[Manedwolf]

But the major reason for developing the Model III was because the FCC had just instituted new regulations about RF emissions generated by computers and other electronic devices. The Model I was completely unshielded and was unable to pass the emission restrictions.

 

[mtnbkr]

 

Why are you ?  It wasn't a safety issue, just an interference issue.  I think they're talking about Part 15, which refers to interference with external devices.  Many electronic devices fall under these regulations.  It's an RF pollution issue, not health and safety.

Chris

[Manedwolf]

Why are you ?  It wasn't a safety issue, just an interference issue.  I think they're talking about Part 15, which refers to interference with external devices.  Many electronic devices fall under these regulations.  It's an RF pollution issue, not health and safety.

Chris

They didn't leak excessively from an unshielded transformer or something? I don't know. I figured they leaked a lot of EM or something?

Also, I remember these, that they gave out at the same time in the schools...

[mtnbkr]

I don't know exactly how much they were radiating, but I doubt it was enough RF to be a safety issue.  More than likely, the problem was the RF from those unshielded computers interfering with other electronic devices (TVs, etc).

Chris

[Nitrogen]

They didn't leak excessively from an unshielded transformer or something? I don't know. I figured they leaked a lot of EM or something?

Also, I remember these, that they gave out at the same time in the schools...

Oh man, I remember that one.  Superman gets stupified somehow, and a bunch of kids on TRS-80's have to do calculations for him.  It was totally stupid and awesome at the same time.

[Manedwolf]

I don't know exactly how much they were radiating, but I doubt it was enough RF to be a safety issue.  More than likely, the problem was the RF from those unshielded computers interfering with other electronic devices (TVs, etc).

Chris

Ahh, okay. I was thinking of something in terms of those "shoe store" x-ray machines that blasted everyone nearby with no shielding, but I guess the computer wasn't anywhere near a high-energy device.

As far as EM, I'd found a scary little toy in an old bin once. It's a "bulk tape eraser". It heats up fast and is only supposed to be turned on for a moment, but as soon as you trigger it, any CRTs even in the same room immediately get a distorted picture.

[RaspberrySurprise]

Ahh, okay. I was thinking of something in terms of those "shoe store" x-ray machines that blasted everyone nearby with no shielding, but I guess the computer wasn't anywhere near a high-energy device.

As far as EM, I'd found a scary little toy in an old bin once. It's a "bulk tape eraser". It heats up fast and is only supposed to be turned on for a moment, but as soon as you trigger it, any CRTs even in the same room immediately get a distorted picture.

That's quite the WMD you've got there.

[Paragon]

I just wanted to say thanks everyone, for making me feel really really young.   

[Bogie]

I have a Kaypro 4 in the storage unit. It worked when I plugged it in last.

Wordstar rawks.

[ArfinGreebly]

Kaypro 4 and two Kaypro 10s.

All worked last time I fired them up.

I forget when that was, though.

Cut my teeth on WordStar.  The first text editor I ever wrote was written to mimic the WordStar keyboard commands and screen behaviors.