Estimated reading time: 6–7 minutes
“- Can you read dumps?”
Immediately after I was hired, my laboratory head was Mr. Izea Feinstein — a walking encyclopaedia — assisted by Mr. Schechter, a man of about fifty. One of them dealt with cost-price analysis and tariffs in railway transport, the other with the same thing but in road transport. We made a good team: I learned from them the cost structures in transport and the methodologies for devising tariffs; they tried to learn programming from me. A fair exchange: they gave me the reality; I gave them the tools. I do not know how much programming they learned from me, nor how much use it was to them wherever they emigrated, but I can say with certainty that the cost-pricing structures and tariff calculation methodologies I learned from them served me in full.
I used programming to carry out the tasks I was given: a daunting volume of complex calculations, data series, statistics, scenarios, “what if…” analyses — all fed through punched cards as the data input medium and returned, inevitably, on paper, as results. Today we call it a “pipeline”; back then we called it “the listings have arrived again.”
ICPTT was a large institute, with many sections spanning the domains of railway, road, and naval transport, roads and bridges, occupational psychology, economics — everything that moved (or, rather, everything that ought to have moved). In the technical research sections and laboratories, most of the work was done in FORTRAN, while in the Economics Section the language of choice was COBOL.
I worked fluently in both. It was January 1972.
As I have said, we ran our programs at the Computing Centre of the Romanian Railways, on a Siemens computer. And here comes the part that, without my realising it at the time, changed my standing within the institute. I quickly discovered that the Siemens “at the Station” — as we called the CFR computing centre — was built on an architecture compatible with the IBM System/360 family. That explained why so many things “sounded familiar” to me from the outset: the logic of the registers, the discipline of the addresses, the way an error looked when it decided to be terminal. (The Siemens System 4004 was, in fact, described as System/360-compatible.)
At the institute, the breakthrough I made was with reading dumps.
For those who have not had the (dis)pleasure: a dump was, in short, the autopsy of the computer. When a program collapsed or ran into a serious error — to put it more plainly, when it “crashed” — the system would produce a listing, sometimes kilometres long, containing a photograph of the memory and the registers at the exact moment of the accident. Everything in addresses and values, expressed in hexadecimal — that is, a convenient representation of the binary contents of the memory and the registers. If you did not know how to read it, a dump was nothing but a curtain of digits, like rain on a tram window: a great deal of information, zero meaning. If you did know, it was like a detective novel: you could see where the film had snapped, who had pulled the trigger, which operand had encountered another operand in the wrong place. Today the same concept lives on more elegantly: “core dump,” “crash dump,” “memory dump,” plus logs, stack traces, debug symbols. The difference is that nowadays the dump no longer arrives in reams of paper, reeking of ink and frayed nerves. It arrives on screen. And, if you are lucky, with a polite message.
A modern programmer uses tools such as WinDbg or GDB to “navigate” through that hexadecimal ocean, much as his predecessors’ traced addresses on paper. But when all is said and done… it is still a dump — only you can no longer use it as a doorstop.
At present — 2026 — an even more modern programmer uses an AI model, more modern still, to generate code that works “first time round.” But let us not get ahead of ourselves. We are still at the beginning of 1972.
One day, a researcher asked me from the doorway, in an almost indifferent tone:
— Can you read dumps?
The question was posed as casually as if he had asked whether I knew how to tie my shoelaces. I answered cautiously:
— I might.
Within a few minutes, a voluminous dump had landed on my desk. The format was familiar, but I asked for a reprieve until the next day. Not out of modesty. Out of a survival instinct. I pulled out my notes from university, retraced a few paths in my mind, and yes: I was on familiar ground.