Yeah, it was all tapes. We only had to use them once when I worked there: after finding out the UPS connected to the mainframe was a dud. And then it really was roulette because the first two tapes were unreadable, so we ended up with three week old data.

I’d believe it’s real. In 2016 I was at a company trying to migrate off an old IBM mainframe and green screens. It wasn’t like an airline with complex or critical code; it was just a barely functional ERP for a warehouse. Source control was the furthest thing from their minds. Some companies and IT departments are very reluctant to change, regardless of how much time and money it save.

It gets worse if you use Microsoft D365 AX products. Then you have to provision an entire Build server for builds which has to run Visual Studio 2019 on Windows 10. To do a build you run a pipeline in Azure DevOps, which runs the compiler in a full Visual Studio 2019 environment, which has to run on a special Azure virtual environment running Windows 10 hosted by Microsoft. It’s so fragile.

deleted by creator

I used to write extensively with C++, but it has been a long time since speed mattered that much to one of my applications. I still marvel at the cache-level optimizations some people come up with, but I’m in the same mindset as you now.

My workload split of Data Movement vs Data Transformation is like 95:5 these days, which means almost all the optimizations I do are changing batch/cache/page/filter settings. I can do that in any language with http bindings, so I choose those that are faster to write.

My cheap old 3D printer requires constant fiddling before and after every print, yet still fails probably half the time. I avoid printing things sometimes just because I don’t want to deal with it.

I would still agree with you 100%. I hate my HP printer so much.