ICL 1900 Series Computers

ICL 1904A in 1981, photograph courtesy Phil Chipper

It has been suggested to me that I include a section of reminiscences, with names removed to protect the guilty. If you want to contribute, please email them to me.

I will start things off with a few memories of my own.

• I joined one organisation as the Systems Programmer to set up and support George 3, on the soon to be installed 1904S, when the builders had finished the new purpose built computer centre. After the system was up and running, we ran a test for the backup generator that powered the computer suite and found:
  1. There were no 'battery maintained' lights in the underground computer room - it went extremely dark when the power failed until the emergency backup (from another system) cut in.
  2. Our generator started automatically, but needed manual switchover, which involved going out to the plant room, opening a power distribution cabinet, putting your hand in and throwing the switch (at which point the plant room lights came on) and then being able to see to take your hand out from the live 3-phase (415/240 volt) bus-bars.
  3. The generator had been set up to vary its frequency in order to maintain a stable voltage, which played havoc with the EDS60 drives as they used synchronous motors (locked to the 50Hz mains frequency) to run the discs at the correct speed.
  4. The computer room got hotter and hotter while the air conditioning plant was idling - it was found that the chart recorder/sensor panel hadn't been connected to an 'essential supply' circuit; not all was lost however, the coffee machine was still working.
     
     
• On another occasion, an outside (dial-in MOP) user had written a macro to RETRIEVE (to online filestore from offline tape storage) all of the very many files in his directory, which he issued before going home for the evening. The very evening that we ran the weekly payroll for hundreds of employees. Needless to say this caused a few problems, the system requests to load DUMP tapes taking priority over the production payroll work. In the end, we had to run down the workload and re-boot the system to cancel the retrieves, after which I ran CHUCKOFF (an added ARCHIVE NOW command) for the whole directory and changed his password to 12 'backarrow' characters. For those not familiar with the 1900, 'backarrow' on a teletype acted as a backspace. The next day, when he couldn't log in, I gave him the new password (without pointing out that it couldn't be typed in) and let my manager have a few words about sensible use of the system.
  
  
• Another payroll story. One evening I was at our social club playing badminton, as were most of the computer centre staff - it was our regular weekly get together. The phone behind the bar rang (mobile phones in those days?), and I was told that a taxi was on its way to pick me up - the system had crashed during the weekly payroll run. I got in, still in badminton kit with a track suit over the top (I lived 10 minutes walk away from the club and went down in my kit), and it turned out to be that the online filestore was full and the unjammer had crashed. We had to do a 'general restore', not a common occurance, but one to be avoided especially on a payroll day. Once the restore was under way, one of the operators loaned me a towel and I got a quick shower - some of the technical details were lacking, but not staff comfort! The software team leader arrived a bit later and the two of us finished getting the vital payroll runs through by the early hours of the morning, the operators having gone home at their normal 'end of shift' time. When I arrived later that morning, an ambulance, with its blue light flashing, was waiting outside the computer centre ready to distribute the documents (payslips, cash breakdowns etc.) to the various pay offices.
  
  
• On one occasion, the monthly accounting run (punch data on KeyEdit system, read in card images from tape, process, print and decollate output) was completed and the output split and sent to the various user departments, when a few hesitant phone calls came in querying the output sent to the various Accounts Offices as the prevoius month's items appeared to be missing. It turned out that the first job in the suite (read in the data) had been omitted, thus the print was a duplicate of the previous month. The Chief Operator's response was "its only 40 boxes of paper and a bit of overtime for the lads on Saturday" - guess who forgot to run that first job.
  
  
• I was with a team running some operating system tests at a clients' site, using their old machine. We needed a power point to plug in some VDUs, when a colleague suggested plugging in under the false floor where a bank of tape decks used to be. We duly lifted a tile and found water lapping around the power connectors (415/240 volt 3-phase), needless to say we found the 'emergency stop' in a hurry. We were later told "not to worry, its just the overflow from the air conditioning, only worry when it starts coming out through the tiles".
  
  
• From Phil Chipper:-
  The stories I have tend to be more of sense of how it was to be an operator at the time, for example:
  • It was not uncommon to play pranks on new operators, particularly during evening/night shifts. Typical examples being:
    • Sending a trainee to the on site engineers in their workshop for a "Long Weight". Inevitably what the unfortunate individual got was a "long wait", before being let in on the joke.
    • Requesting parity bits to top up the hopper in the back of the processor was another, as was a bucket of steam to maintain the level of humidity in the machine room.
      
      
  • Some times the pranks got less subtle. One particular shift leader, who tended to leave a great deal of the work to his more junior operating staff, finally exhausted their patience and was found by the shift manager peering up through one of the air vents in the false floor where he had been left trussed up like a chicken.
    
    
  • The Write Permit Rings (WPRs) in the back of the tapes, which worked like the tabs in the back of cassettes, were often used as ammunition in fights that broke up the boredom of a lengthy tape intensive job or possibly the last dump before closedown. Although care needed to be exercised as a heavily launched WPR could "knock off" a deck and cause the run to fail, which in the latter case would delay the shift's departure and leave the unfortunately individual regretting his equally unfortunate accuracy.
    
    
  • Getting married could also be traumatic as one operator during a night shift was taken, wearing just his normal clothes, stripped of all id, cash, etc., then bundled into a car and left several miles from the data centre. This was during the middle of a very cold winter, not really the time for stroll in your shirt sleeves.
    
    
  • Operators were also generally regarded as cheap labour/cannon fodder by the management and got stuck with lots of unpopular and strange jobs, e.g.
    • Hosing down the flat machine room roof during a particularly hot summer as the air conditioning couldn't cope and the machines were in danger of overheating.
    • Humping, and delivering any/all of the incoming/out going items you'd expect in a data centre.
      
      
  • One other job was switching the alternators. Before the advent of UPS the installation had two alternators which were used to take the spikes out of the "dirty" incoming mains current. One was in use the other as standby. Periodically they were swapped to even wear ensure they were working. The basic processes was to run the spare up to speed and when the current was stable switch to it, then power down the one that had been in use. Simple process but it involved quite a number of actual steps to achieve and inevitably someone forgot to make the switch before starting the power down with the result that all the machines ground to a halt. Curiously enough some years later when these alternators had been replaced by a UPS system the same thing happening during its maintenance when the engineer cut the power before switching in the override to the UPS generator.
    Both the above happened during the day. Unigate's data centre although having a modern machine room was based in an old house (Bellefield), which was reputedly haunted. On one occasion during the time when we converting from the ICL to Sperry machines when the machine room had effectively four mainframes and their peripherals in it and before the UPS system was installed, we had a power-cut in the middle of a storm. In an instant we went from a brightly lit noisy environment to almost blackout (emergency lights aren't up to much). The only noise was the sound of the disks winding down and the wind outside, very spooky.
    
    
• From Martin Taylor:-
  When I joined the dreaded Pergamon Press in 1976 the machine room held (ISTR) a 1902T and a 1903A. The story goes that the boss, the much-unlamented Robert Maxwell, brought a business associate into the room one day (well, it was his machine room, he could do what he liked!), and, waving his hand towards the row of tape decks, declaimed, in his distinctive gravelly voice, "...and these are our random access devices!". Nobody corrected him.
  
  
• From Martin Taylor:-
  A colleague in ICL's Software Distribution Department at Friar Street, Reading, wrote a game of tic-tac-toe entirely in George 3 JCL - no compiled program whatever - seriously abusing George 3's SETPARAM command. I suspect he went on to become a better programmer than I ever was.
  (Has anybody got a copy of this? - BWS)
  
  
• From Martin Taylor:-
  When I joined ICL I knew practically nothing about computers (as opposed to today, when I... oh well). I remember my astonishment when someone pressed a key on a console teletype in the Cardiff Road machine room and several yards away a line printer clattered into action. It was gently explained to me about cabling under false floors.
  
  
• From Martin Taylor:-
  Another SDD colleague, who I shall call Rodney, was a guy with a bit of a problem, and prone to an exaggerated view of his own competence. A friend and I spent a good deal of effort (and trained ourselves in parameter validation techniques while we were about it) trying to make our operational macros "Rodney-proof". Sadly, we never did succeed, which was actually part of the learning process in itself.
  
  
• From Knobby Clark:-
  The development of G1/G2 used magnetic tapes to store the "common source". There was a nightly "edit run" which read the edit cards, modified the source and generated the next tape in the cycle. There were seven tapes in the cycle so that after a week you effectively had no backup (OK we know better now but this was 1968). One night, the junior programmer told me that "the edit run had failed". I assumed that (let's say "he") "he" would put a new tape into the cycle...........but no "he" just changed the run to edit from A to C instead of from A to B. And the next night, the edit run "Failed" again. And so without telling me "he" changed it to edit from A to D, ........ and the next night "he" changed it to edit from A to E. Fortunately, at this point, I decided to take a more detailed interest. Of the course, the problem was that tape A was faulty...........so we started again from the tape F which "he" was just about to overwrite.
  
  
• From Knobby Clark:-
  For some years the common source contained a file called NELLIE that contained a lovingly transcibed copy of the text of "Eskimo Nell". One one occasion the source tape was supplied to an academic institution, presumably to assist their operating system researches. Tongue in cheek, they sent a letter thanking us for the text and pointing out the obvious spelling mistake in the third stanza. Unfortunately the letter was sent about three levels too high in the ICT hierarchy which prompted a stern internal memo headed "Obscene Poetry". And that's why the subsequent copies of the source tape contained a file of unreadable characters. If you have a copy of the source tape, try shifting each character of that file one bit to the left or right, I can't remember which.
  (Unfortunately, I don't have a copy of the G2 Source Tape, but I do have a copy of #NELL, a program which prints this text, complete with spelling error - BWS)
  
  
• From Knobby Clark:-
  In the late 1960's the punchroom at ICT had a turnround time of several days so we used to punch our own cards. We had no access to an interpreting punch; we used the blind hand punches. As you imagine, this was a fruitful source of errors so we adopted the practice of keeping cards that had been through a successful edit run sorted by their content.........*ALT 0,1 followed by *ALT 3,3 etc. We also kept the GIN instruction lines so we had lots of

                        LDN   0  0
                     

  cards and lots of

                        EXIT  0  0
                     

  cards and so on. And the result was that on one occasion I was trying to debug 250 odd lines of tightly written GIN code which had only one commented line.

                        LDN   0  0          [CLEAR X0