Interviewer: Tell me an interesting debugging story
Interviewee: …
Great documentary on the Voyager team: It’s quieter in the twilight
Still faster than the average Windows update.
Windows 13 update log:
Change kernel to Linux.
Build custom OS for astrophysics and space science applications.
happy rocket engineer noises
Microsoft can’t even release a fix for Window’s recovery partition being too small to stage updates. I had to do it myself, fucking amateurs.
Can’t or won’t? The same issue exists for both windows 10 and 11, but they haven’t closed the ticket for windows 11… Typical bullshit. It’s not exactly planned obsolescence, but when a bug comes up like that they’re just gonna grab the opportunity to go “sry impossible, plz buy new products”
I didn’t know that. So the ticket is still open for 11 but there’s still no fix?
Not to mention what a bitch that partition is when you need to shrink or increase the size of your windows partition. If you need to upgrade your storage, or resize to partition to make room for other operating systems, you have to follow like 20 steps of voodoo magic commands to do it.
Whoa learned that one at the weekend. Added a new nvme drive, cloned the old drive. I wanted to expand my linux partition, but it was at the start of the drive. So shifted all the windows stuff to the end and grew the Linux partition.
Thought I’d boot into windows to make sure it was okay, just in case (even though I’ve apparently not booted it in 3 years). BSOD. 2-3hrs later it was working again, I’m still not sure what fixed it of I’m honest, I seemed to just rerun the same bootrec commands and repair startup multiple times, but it works now, so yay!
Well, they only had to test it for a single hardware deployment. Windows has to be tested for millions if not billions of deployments. Say what you want, but Microsoft testers are god like.
I still cannot believe NASA managed to re-establish a connection with Voyager 1.
That scene from The Martian where JPL had a hardware copy of Pathfinder on Earth? That’s not apocryphal. NASA keeps a lot of engineering models around for a variety of purposes including this sort of hardware troubleshooting.
It’s a practice they started after Voyager. They shot that patch off into space based off of old documentation, blueprints, and internal memos.
Imagine scrolling back in the Slack chat 50 years to find that one thing someone said about how the chip bypass worked.
IBM is 100, but the Internet didn’t exist in 1924, so we’ll say the clock starts in 1989. I’m pretty sure at least MS or IBM will be around in 15 years.
This is why slack is bullshit. And discord. We should all go back to email. It can be stored and archived and organized and get off my lawn.
I realize the Voyager project may not be super well funded today (how is it funded, just general NASA funds now?), just wondering what they have hardware-wise (or ever had). Certainly the Voyager system had to have precursors (versions)?
Or do they have a simulator of it today - we’re talking about early 70’s hardware, should be fairly straightforward to replicate in software? Perhaps some independent geeks have done this for fun? (I’ve read of some old hardware such as 8088 being replicated in software because some geeks just like doing things like that).
I have no idea how NASA functions with old projects like this, and I’m surely not saying I have better ideas - they’ve probably thought of a million more ways to validate what they’re doing.
You sure? The smell off some of the corpses will have been terrible.
I’m not saying they’re all dead, but an intern at the time of launch would now be 70. Anybody who actually designed anything is… Well… The odds of them still being around are low.
100% they’ve got an emulator, they’ve had dedicated test environments since the moon landing for emulating disaster recovery scenarios since the moon landings, they’ve likely got at least one functioning hardware replica and very likely can spin up a hardware emulation as a virtual machine at will.
Source: I made this up, but I have a good understanding of systems admin and have a interest in space stuff so I’m pretty confident they would have this stuff at bare minimum
That’s my assumption too, but we’re talking about a different era, and I really have no idea how they approached validation and test/troubleshooting.
I’ve seen some test environments for manned missions, but that’s really for humans to validate what they’re doing.
V’ger was quick 'n dirty by comparison (with no criticism of the process or folks involved…they had one chance to get these missions out there).
There is an fascinating documentary about the team that sends the commands to Voyager 1 and 2 called It’s Quieter in the Twilight
To add to the metal, the blueprints include the blueprints for the processor.
https://hackaday.com/2024/05/06/the-computers-of-voyager/
They don’t use a microprocessor like anything today would, but a pile of chips that provide things like logic gates and counters. A grown up version of https://gigatron.io/
That means “written in assembly” means “written in a bespoke assembly dialect that we maybe didn’t document very well, or the hardware it ran on, which was also bespoke”.
They also released the source code of the Apollo 11 guidance computer. So if you want to fly to the moon, here is one part of how to do it.
I was already impressed when they managed to diagnose a single bit flip a few years ago.
