Your comment makes me think of how they packed Crash Bandicoot into memory. [1]

[1] https://www.quora.com/How-did-game-developers-pack-entire-ga...

Jesus Christ, 4MB? Did they write the whole thing in machine code?

Back in the day, not everything needed jquery and bootstrap deps to be considered "modern" ;)

Jokes aside, russian space program had some AMAZING developers working in some very low-level languages with unbelievable efficiency. I had the pleasure of working with some of them many years later on enterprise software at large banks.

NASA has written some amazing software as well. Start here: https://news.ycombinator.com/item?id=8063192

You may joke, but Node.js is still going strong after six years, while Buran never went past test flights. Imagine how much farther Buran could have gotten if they had the ability to share small, focused, reusable modules with the community. Imagine how much more reliable their systems could have been, if they had been able to reason about them at a higher level of abstraction, without getting bogged down into the low-level complexity of left-pad!

I know you are being sarcastic, but remember that there was no community at that time. And besides, open source code would have meant that other countries would have gotten it.

Also, check this out: http://www.fastcompany.com/28121/they-write-right-stuff

Very different from how a typical Node.js left pad module is implemented.

I get your meaning, but additional abstraction layers are equally dangerous because they obsfucate underlying constraints and result in incorrect reasoning about a system.

There's a reason RTOSs don't have very tall software stacks with multiple layers and virtual machines between them and the hardware.

When your garbage collector invoking at the wrong millisecond 0.01% of the time means everyone on the craft is incinerated in the atmosphere... that's a different level of reliability requirement.

You missed the sarcasm.

If you have some stories you can/would like to share, please do!

4MB is rather a lot for embedded applications, especially if they are written in assembler. I can't imagine having to hand-check 4MB of assembler for flight certification.

Remember we had an entire game industry, graphical assets and all, distributed on 1.44MB floppy disks.

The Apollo Guidance Computer had about 64KB of memory. The first minicomputer that I was a product manager for in the mid-eighties had a max memory of 4MB.

(In the latter case, not everything was written in assembler but it was all compiled.)

Given the time probably was all assembly, at the time four MB is quite a lot.

Early on, many space data systems used wire core memory to reduce errors.

I'm wondering whether they built a simulation environment to test their software.