How did people learn to program before 1991 without the Internet? If you got stuck, how did you get past it?

Quite simply: you had to think for yourself at the time 🙂

By the way, many people in Germany only had internet from 1997/98, when we were spams with AOL CDs.

So what was it like back then in the dinosaur era?

There were books.And if you grew up like me in an area where you were the only programmer far and wide, there was no one to question. I printed the program code of Borland’s TurboVision (an application framework for Turbo Pascal) in 1992 and spent weeks studying line by line. Some lines I only really understood after many days or weeks: virtual methods, for example. I couldn’t ask anyone what that is. The exact understanding of this occurred to me at some point in the bathtub.

But if i had an error in 1994 and was accessed writing on an unsecured pointer, for example, or allied memory was not fully released, the entire Windows 3.11 simply crashed. The unsaved work was gone.And the required new boot on my 486 took up to 10 minutes.

This time was very costable at the end of the day, because it just happened several times.So you sit down and think and try to understand everything that’s going on. Because back then, when the operating system crash happened, I wasn’t even aware that I had accessed a wrong pointer. there was a wrong value in my pointer. I had to learn that first. I was always going for a walk at the time and just thinking. Sometimes I tried around and then suddenly understood the problem.

I often went to bed frustrated with the problem and woke up in the morning with the solution.The brain processed things during sleep and developed an understanding of what was going on.

Today, such questions or problems are usually answered with another tab in the browser on StackOverflow in seconds.So you let others think for themselves and just take over the solution. I believe, however, that the approach described above provides more foundation: to think for yourself and to work out the solution itself (or in groups). However, this again takes time…

The answer to this question is herewith.A few more thoughts:

When we talk “old bunnies” with younger developers, who nowadays have grown up with script languages and garbage collectors (not all of them, of course, but mostly) and sit at a very high level of abstraction, which they use through Java, PHP or C. Net-framework, of course garnished with StackOverflow and Quora, often result in very interesting conversation constellations:

We old people have all worked with Assembler or C or Pascal, we “think” like a CPU, we know what is happening, we still pay attention to things that today sometimes have no value (save memory, save CPU cycles etc), we still have our own in 1990 Sort algorithms (Bubblesort, Quicksort, etc.) instead of simply sending an “ORDER BY name” to the SQL server, and the younger ones then jokingly say “Oh, Grandpa tells about war again” 🙂

Conversely, I often notice that the younger ones are more concerned about clever architectures and design patterns, which were not yet sooo popular in the “old” world with us dinosaurs.

But we also didn’t have any fat node_modulesfolders in our luggage, where a simple importstatement is enough to get a wealth of libraries and modules.We have mostly always written all this ourselves and were therefore much slower than is the case today with the modern tools, languages and frameworks!

Nevertheless, the old principles, i.e. the rules of the old times, are becoming popular and necessary time and again:

In today’s IoT world, as well as in the blockchain world, low memory consumption and low CPU cycles again play a very big role.Both cost money, for blockchains such as Ethereum, for example, as part of the fee you pay per transaction and per deployment of a SmartContract. And in the IoT world, small, slow hardware with low command sets and low memory is used again.

Even with home automation or in a navigation system, an unbelievable amount of data has to be stored very sparingly and efficiently due to the limitation of the hardware.

And even in eCommerce (at least 5 years ago it was still like that) whether your TimeStamp in the Analytics table was a DateTime or a Unix timestamp per 32bit integer, once you add 120 million rows per month.

We old hares still know all this and are in fact robbed with loincloth through the jungle.There were traps everywhere, and we took every trap with us 🙂 However, we have learned from every single trap and I think that is very important. Nevertheless, I am glad that today we have to sit on a completely different pedestal in modern times and not everyone has to reinvent every wheel again and again!

Leave a Reply