I just spent forty-five minutes of my life tracking down a point of early computer programming and feel very satisfied with myself.
I was made aware of Victor Gijsbers' appeal for help deciphering a commemorative plate that belonged to his grandfather, the Dutch computing pioneer Bram Jan Loopstra. As far as I could see in the thread on Mastodon, everything on the plate had been identified (date in binary, motto of Simon Stevin, electrical diagram symbol for a pentode, hysteresis loop) except for the sequences of gridded black dots, generally agreed to be a 5-bit code, assumed to be instruction code for ARMAC or one of its predecessors at the Mathematisch Centrum in Amsterdam. I wondered immediately if it was a piece of the program for Dijkstra's shortest path algorithm, which was famously used to demonstrate the capabilities of ARMAC in 1956, but then I had no idea what the original code looked like. I didn't even know what language would have been used to program an early Dutch computer in the mid-1950's. Thereupon I spent a half-hour butting down blind alleys of early programming languages and autocodes before finally hitting on an archive of the manuscripts of Edsger Wybe Dijkstra, after which I had a quick answer:
My only model was the program organization for the EDSAC in Cambridge; I followed it closely when designing program notation, input, output and library organisation for the ARRA in Amsterdam. For the next machines, the FERTA, the ARMAC and the X1, program notation and input would very much follow the same pattern: I clearly was a conservative programmer. Add to this that the ARMAC's instruction buffer with a capacity of one track of the drum, destroyed the store's homogeneity, and you will understand that I did not embark on adventures like "autocoders".
If the dots on Gijsbers' plate were displayed by any of the above-named computers that Loopstra et al. built and Dijkstra programmed, then they should be legible to anyone conversant with the programming of EDSAC. The question was whether its programming techniques were still readily available; the computer was famous enough that I guessed they might be: they are. I am now seriously thinking about downloading an EDSAC simulator. I used to know some FORTRAN. It looks fun.
I don't expect to be able to translate Gijsbers' plate. If it is commemorating some project other than ARMAC, the code is even less likely to be a quotation from Dijkstra's algorithm, although if I hadn’t been trying to find out what languages he used, I don't know how long it would have taken me to stumble across the connection with EDSAC. Frankly I am assuming that someone better versed than myself in assembly code has already figured it out, I just haven't seen it because it didn't happen in that thread and I'm not on Mastodon. But still: I was starting from zero where this machine was concerned and now I know something about it. One could, of course, use plasmodial slime mold to solve the shortest path problem, but in 1956 no one knew that.
I was made aware of Victor Gijsbers' appeal for help deciphering a commemorative plate that belonged to his grandfather, the Dutch computing pioneer Bram Jan Loopstra. As far as I could see in the thread on Mastodon, everything on the plate had been identified (date in binary, motto of Simon Stevin, electrical diagram symbol for a pentode, hysteresis loop) except for the sequences of gridded black dots, generally agreed to be a 5-bit code, assumed to be instruction code for ARMAC or one of its predecessors at the Mathematisch Centrum in Amsterdam. I wondered immediately if it was a piece of the program for Dijkstra's shortest path algorithm, which was famously used to demonstrate the capabilities of ARMAC in 1956, but then I had no idea what the original code looked like. I didn't even know what language would have been used to program an early Dutch computer in the mid-1950's. Thereupon I spent a half-hour butting down blind alleys of early programming languages and autocodes before finally hitting on an archive of the manuscripts of Edsger Wybe Dijkstra, after which I had a quick answer:
My only model was the program organization for the EDSAC in Cambridge; I followed it closely when designing program notation, input, output and library organisation for the ARRA in Amsterdam. For the next machines, the FERTA, the ARMAC and the X1, program notation and input would very much follow the same pattern: I clearly was a conservative programmer. Add to this that the ARMAC's instruction buffer with a capacity of one track of the drum, destroyed the store's homogeneity, and you will understand that I did not embark on adventures like "autocoders".
If the dots on Gijsbers' plate were displayed by any of the above-named computers that Loopstra et al. built and Dijkstra programmed, then they should be legible to anyone conversant with the programming of EDSAC. The question was whether its programming techniques were still readily available; the computer was famous enough that I guessed they might be: they are. I am now seriously thinking about downloading an EDSAC simulator. I used to know some FORTRAN. It looks fun.
I don't expect to be able to translate Gijsbers' plate. If it is commemorating some project other than ARMAC, the code is even less likely to be a quotation from Dijkstra's algorithm, although if I hadn’t been trying to find out what languages he used, I don't know how long it would have taken me to stumble across the connection with EDSAC. Frankly I am assuming that someone better versed than myself in assembly code has already figured it out, I just haven't seen it because it didn't happen in that thread and I'm not on Mastodon. But still: I was starting from zero where this machine was concerned and now I know something about it. One could, of course, use plasmodial slime mold to solve the shortest path problem, but in 1956 no one knew that.
![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}