Computing

My first computer was an IBM 1620 at the Crystallography lab. of Prof. Robert Gay (U. Bordeaux I, Talence, France).

My first program (a few tens of punched cards) was written in FORTRAN and the two-pass compiler was about two full boxes of punched cards. The compiling process was has follow :

1. place in the card reader the first pass compiler plus the manually punched cards of your own program
2. compile (many lights blinking for a while)
3. wait for the lengthy punching of 5 to ten times more cards than the original program
4. read the error message on the typewriter. If any, trash all the punched cards, correct you code and go back to step 1
5. append cards punched at step 3 to the second pass compiler
6. compile again (many lights blinking for a while)
7. wait for the very lengthy printing of even more cards than at step 3
8. read the error message on the typewriter. If any, trash all the punched cards, correct you code and go back to step 1

On the photos, note the printer, a typebar mechanical typewriter, a noisy, very slow and unreliable input/output device.
Also note the bulky punched card reader/writer (2 or 3 cards per second!)
These photos does not show the huge cabinets full of ferrite rings and wires: the 40 k decimal memory!

Then I used CDC 3600, IBM 360 and CII Iris 80 mainframes. In 1970, I could even admire an IBM Stretch (IBM 7030) at the Limeil-Brévannes CEA centre.

The maintenance console had millions of small pulsating, flickering bubbles with sudden overall busts like flash of lightning, sun dawn, etc.
Operators were constantly mounting and dismounting tons of magnetic tapes.
As many other people I sepnt a lot of time admiring gorgeous spectacle.

My neutron diffractometer D8, and 5 other ILL diffractometers, were simultaneously, and slowly, driven by a French computer, Télémécanique Electrique T2000 with the following characteristics:

• T2000 (Télémécanique Electrique) 2x32 kWords, 19 bits, 1.5 µs per cycle
• 2 fixed head hard drives, 256 kWords, 10 ms
• 2 mag tapes, 9 tracks, 800 Bpi
• 8 typewriter, 10 char/s
• 1 tape puncher/reader, 50/110 char/s
• 1 card puncher/reader, 400 cards/minute
• 1 CRT screen with keyboard

The programs were written in a specially designed realtime FORTRAN II (yes "real-time"). Everything was slow, the disk space was very tight (only 1 Mb shared by 6 instruments!). I spent many days and nights on that computer at programming and fighting with the instrument. 50% of the beam time was lost due to computer latency. See here for more details (pdf - 1.99 Mi).

At the ILL, I discovered "modern" computing with the mainframe DEC 10 (also named PDP-10, DEC 1007 System 10) installed in 1974 and modernized in 1981. This 36 bits computer offered interactive programming, interactive graphics screens (Tektronix 4010, DEC GT40), removable hard drives, etc.

The DEC10 was by far more modern than IBM or CDC computers. The OS commands were extremely simple and intuitive, with such a simple syntax that there was no real need of a manual since you could easily figure out commands. The command sets of PDP11 (RSX11M or RT11) and VAX (DOS) computers were similar but less intuitive, however by far more user friendly that the incredibly cryptic and hard to memorize terminal commands of Unix.

For example, any terminal (terminal printer, card reader/puncher, printer, disk, magnetic tape, Dec tape, punched tape, ...) was treated the same through commands with the same syntax and one could copy files from one terminal to any other using one single command. Doing the same on IBM computers was a nightmare and an orgy of JCL punched cards like the famous "SYSIN DD".

The use of that computer was so simple that one hour of training was enough to get granted the "driving licence" and the keys of that mainframe.
Yvon siret told me how to implement dynamical arrays in my FORTRAN IV programs. Something quite unusual at that time.

I really loved that computer ! See here some documents about the Dec10 at the ILL.

DEC1091S characteristics: KL10E processor, 1 million words (36 bits) of memory, 2300 MBytes of disk storage, 5 magnetic tapes units and a communication front-end, including support for up to 64 terminals.

I also had a lot of fun with the first interactive game, the fantastic "Lunar lander" or "Moon lander" (1973) on the terrific DEC GT 40 graphical terminal!

However most of the work was done on Tektronix 4010, a slow monochrome graphical terminal but offering basic interactivity. The main problem at the beginning was the lack of graphical printer. We used Polaroïd cameras instead but the photos were small and with a low resolution. When they finally arrived, printers used a thermal paper that rapidly turned dark brown at sun light.

Then I worked on PDP 11 and VAX computers both excellent DEC products. I mainly used the FORTRAN language for scientific programming. This did not preclude fun since I played with global sections (shared memory), master/slave applications for instrument control and data processing, memory page management, etc. I also implemented plugins in some of my FORTRAN programs through the use of two little known VMS system routines: LIB$FIND_IMAGE_SYMBOL and LIB$CALLG for the dynamical linking of routines.

During years 80, I also fought a lot with "escape sequences" programming. Escape sequences were characters embedded in the data sent to a video text terminal or typewriter. They were used to control formatting, line length (either 72 ou 132 characters) and other output options. Unfortunately enough they were often vendor-specific and even "standardised" ANSI escape sequences did not acted the same on different terminals. Thus I wrote a routine that auto-recognized the many terminal types available at the ILL at that time (video text: T4010/14, GPX, VS200, VT100/102/125/132/200/300, etc. ; typewriters: LA32, LA120 or MINITHERM, etc.)

The years 80 were also the time of large pen plotters (BENSON, CALCOMP) and of GKS (Graphical Kernel System) programming. The later was my first contact with the concept of graphic ports, a good introduction to the graphic ports of the Macintosh toolbox.

I was absolutely convinced of the total supremacy of DEC VAX mainframes until late 1984 when a colleague, Georges Messoumian, showed me an Apple Macintosh 128. At that time I was preparing my PhD and I was very busy making many pen and ink drawings of charts and instrument layouts. I was also developing my own word processor written in Fortran IV (VAX VMS) and capable of bold, underline, subscript and superscript through escape sequence programming for a daisy wheel printer. While the print quality was average I was very proud of the result but I still had to manually write the mathematical characters and glue the drawings on pages.

Thus, just imagine my surprise when I discovered MacWrite, MacPaint, MacDraw and the ImageWriter. Bitmap images, vectorial drawing, texts with several fonts, rich formatting options and all that directly printable on a same document... no expensive mainframe was capable of a tenth of what a Mac could do !

I started Pascal programming on a Mac plus and wrote several scientific applications (ABFfit, ABCstat, MacSurvival, MacXenon, OrientExpress, PkFit, etc.). Most of them did not survive the switch to Mac OS X. However I can still use them on modern Macintoshes through the use of SheepShaver, a Mac Classic emulator.

I am still a loyal supporter of Apple's computers and the current president of the "Macintosh Alpes Club", a developers group which started with the birth of the Macintosh.

My best Macintoshes were :
      1986 - Mac plus with a hard drive
      1990 - Mac II Fx with a 21" color screen. Wouahou !
      1992 - PowerBook 180. I remember people looking at me in the train or bars when I was using it.
      1999 - PowerMac G3, nice and fast.
      and then Mac OS X and Intel processors arrived.
      2001 - PowerBook G4 Titanium. Fast, light, 3.5 h of autonomy, ... the perfection !
      iMacs and MacBook pros are excellent too.