Clearly it an be done, considering Mark I as an example. It's also possible to get into a lot of trouble if you aren't skilled enough. A beautiful example is the ARRA computer designed in Amsterdam around 1950. It was a relay machine, asynchronous (no clock; the designers were theoretical physicists who didnt know about such things). Perhaps the worst mistake was that they used relays that for some reason had two coils, and figured they could use that to treat the relay as an OR gate. In the end, they got it to work well enough to run a program once, at the official opening of the lab. To minimize risk they made it a random number generator program. :-) But that was the end of it. So they hired a student of Aiken and built an entirely different machine (with vacuum tube logic, no relays), called ARRA II to make it appear to be just an upgraded ARRA. It's interesting that the designers of ARRA spoke about what they did, and were quite honest about their mistakes. Quite refreshing. Unfortunately that narrative is in Dutch: "Computers ontwerpen, toen". https://ir.cwi.nl/pub/13534/13534D.pdf One of these days I should finish my translation of that lecture. paul

On 2024-07-23 12:09 p.m., Gavin Scott via cctalk wrote: Real .txt files have LF's and CR's.

Very much so. Comparing it with my draft translation I'd say it is quite good, much better than I'm used to from mechanical translations. paul

Yes. And also amazingly poorly trained in practical matters. Admittedly Dutch universities are more narrow than American ones and more likely to emphasize theory; still, it's surprising to come out with a degree in physics and have zero clue about the operation of vacuum tubes. paul

The original is here: https://ir.cwi.nl/pub/13534 and that shows the citation you should use. paul

Dwight wrote: Like minds, I guess. I did the same kind of thing when I was a kid, too. Except, we were pretty poor, and couldn't afford Radio Shack relays. I had to make my own, out of 1/4" plywood scraps for the base, tin snipped from coffee can lids, nails for pivot points, wire-wound nails for electromagnets, and tacks for contacts. I made both usual relays was well as relays with two coils, one to set the relay and another to "unset" it (latching). The wiring was all telephone wire, from a 50-pair cable that my Dad found somewhere and gave to me. My sequential combination lock was for a "burglar alarm" I made for my bedroom door. I made my own keyboards using similar materials, but mine were decimal, encoded with diodes that a neighbor who was into electronics gave to me. The combination was four digits that had to be entered in the correct order on the keyboard outside my door. I also had a keyboard next to my bed that allowed me to override the alarm if someone knocked to come in, as well as to reset the password at any time. The password was stored by sixteen of those set/reset latching relays. The four digit password had to be entered entirely, then an "ENTER" key pressed, and if there was an error anywhere in the password, as soon as ENTER was pressed, the alarm would go off, which was a ringer salvaged from a telephone that someone gave me. The only way to shut off the alarm was to enter the correct code. If someone opened the bedroom door when the alarm was set (by entering the code) the bell would start ringing. The transformer from my train set powered the thing. It worked reliably, but my Mom grew tired of it quickly. Not easy to enter the code with a basket-load of clothes in her arms. I learned a lot from making that thing. The hardest part was making the sequencer that would step as each digit was entered, then compared the entered digit against the corresponding digit of the stored code. That took quite a few relays. I got quite good at making the relays so that they were reliable and consistent in their behavior. It was a total rats nest of wiring on a piece of larger plywood that leaned up against a wall in my room. A few years later, I built a four-bit binary adder using a bit-serial ALU and sequencer made with old telephone relays. The two numbers to be added were entered using a keyboard I made from a salvaged touch tone telephone keyboard. Then, an ADD (#) button was pressed, and the sequencer (which was made with a motor that turned a drum with contacts made from sheet metal screws) would step through selecting each group of two bits, adding them, saving the carry, and then moving onto the next two bits. The result was displayed on five #47 lamps. It was not particularly fast; the fastest I could make it go would give a result in about 1/2 second. Trying to run it any faster resulted in errors, probably due to issues with the design rather than the relays themselves. The thing covered a tabletop, again wired together with telephone wire. I tired of it pretty quickly because it just added numbers. I had visions of somehow making a memory out of relays and making some kind of programming method, again probably using a drum with contacts that would close when a screw was screwed into the drum, such that it could run short programs, but I didn't have the space to build out the rest of it, so, I eventually took it apart so I could have my table back. What I'd give to be able to have all that free time during the summer when school was out. Virtually no real responsibilities (mow the yard, keep my room "clean", do dishes) - just limitless time to dream up projects to build with limited resources. I had no training in logic design, I just kind of figured it out as I went along. -Rick

On Tue, 23 Jul 2024, dwight wrote: Huh? The Z11 was Zuse's first commercially produced machine and it was entirely relay based. It was quite successful and reliable. And this electromechanical memory (or machine) was more or less the _beginning_ of the company, not his down fall. Christian