Home » Mather + Platt » Information systems

Information Systems

In the late 70's and early 80's, the company's commercial and finance systems were based on an IBM 4341 mainframe system. The means of entering data 'en masse' was by the traditional data preparation department who took hand-written paper documents and punched them into the Data General Eclipse 'data-prep' system. This data was then transferred by magnetic tape or a direct connection into the main system. Manufacturing, finance and payroll systems were supported by the IBM 4341 and a team of specialists who developed programs in Cobol and RPG. For interfaces to the operating systems, VM Assembler was frequently used and Nick Lawson was your man.

In the mid-seventies, engineering systems were developed first on a Computer Automation minicomputer. These evolved and were transferred to the IBM 4341 and finally, by the early eighties, to a VAX 11/780 which was better suited to the compute-intensive work, typical in engineering.

The VAX was owned by, and primarily used by, the Rotating Machines Division for its various engineering calculations including programs for vibration analysis and a suite of finite element programs for stress and thermal modelling. The engineering programs were written by mechanical engineers, some of whom eventually transferred to a new department that also managed the computer systems. The programs in use had names like ROTOR, WHIRL, TWIST, HEAT and STRESS.

The Fire Engineering (FE) design department shared the VAX and there was great rivalry between the three FE design engineers David Rowlinson, Philip (Pip) Males and Derek Waistall. For those unfamiliar with the Fortran programming language, the COMMON, REAL and DIMENSION statements are followed by a list of variable names. David preferred to have one variable per line like this;

      COMMON/PIPE/XCOORD(100,250),
     .            YCOORD(100,250),
     .            ZCOORD(100,250)

Whenever David made a change to the programs, he converted the code to his preferred format and whenever Derek and Pip changed the program they converted the code back to their preferred format which was

      COMMON/PIPE/XCOORD(100,250),YCOORD(100,250),ZCOORD(100,250)

There was a rumour at the time that both parties had secretly developed Fortran programs to speed up the Rowlinsonisation and de-Rowlinsonisation processes. Functionally, there is of course no difference between the two formats, but perhaps David was being paid by the line.

This group developed various pipework programs that were intended to automate the design of sprinkler systems. Particular designs were given to the program, which would then calculate all the flows, losses and so on in the system, resize the pipework accordingly and produce a workable design. This program was called Loss Calculation or LC, and this cryptically became ELSIE and the next programs that followed were named HILDA and EMILY. Casual observers were unable to fathom the true origin of these names without assistance.

After a couple of years FE bought their own Prime system. It would have been logical to have continued with the VAX architecture but had this been the case, then the justification for a second similar system would perhaps have been more difficult. However, as things turned out, the Fortran compiler on the Prime was not good at generating code for integer division and gave incorrect results. As a consequence, all Fortran programs had to be checked and all integer divisions had to be converted into floating point. This was a great embarrassment for the team who chose the system. Prime discontinued large systems and concentrated on desktop Unix systems for CAD. Prime was eventually taken over by ComputerVision who themselves were bought by PTC some years later. The Prime logo incidentally was PR1ME, the I replaced by a figure 1.

In those days of the early 1980's, the Digital VT100 reigned supreme. It was one of the first terminals to have a separate keyboard and screen, previous one-piece models being the Lear Siegler ADM-3A, the DEC VT52 and Newbury 7002. The latter, incidentally, had an annoying habit of developing an intermittent fault on the display that could be cured temporarily by lifting the front left-hand corner of the unit by two inches and dropping it back on to the desk. This usually fixed the problem and simultaneously impressed the users and so was quite an achievement. On the rare occasions that this procedure failed, the cause was blamed on the height not being quite the required two inches - a second attempt from a generous distance always did the trick.

The standard VMS screen editor was EDT. In EDT, the F4 key deleted a line but on one occasion when Derek was using the editor, the F4 key stuck down. This in itself might not have been a problem because F1 then F4 would undelete a single line. However, the auto-repeat feature was also enabled and while Derek frantically tried to prise the key up again, his program was gradually getting deleted line by line and disappearing from the screen. The other editors of the day were TECO and SOS (Son of Stop-gap), neither were used to any great extent after the introduction of EDT.

In the early 80's, word processing was in its infancy. PC's had only just been launched and the proprietary systems from DEC and Wang seemed a better solution. Proprietary systems could be developed quickly and without consideration of other emerging standards. Mather and Platt used a Wang VS word processing system and also trialled a DEC Professional personal computer. The Wang was a single central system that was made available to users throughout the business. Printing was still using matrix or daisy-wheel printers, both slow and noisy.

The first IBM PC in the company made its appearance on the desk of Paul Fitton in the central IT department. This had an IBM 3270 co-axial card in it and could emulate a 3278 mainframe terminal.