[TUHS] UNIX testing / error injection

[Adam Koszek via TUHS]

I intuitively assumed there weren't a lot of simulators going on in the early days? When OS itself was a problem, but after the concurrency was there, and perhaps the filesystem was there, I could imagine people writing some C code to simulate disk, network etc. But perhaps I’m wrong? 

In my research on DST there was only YouTube comment about “spin network simulator from Bell Labs”, which may have been an early simulator of some sort — Holtzmann’s first paper on this was 1987. His recollections:

https://scispace.com/pdf/software-verification-at-bell-labs-one-line-of-development-3rj2ypjdo9.pdf

but it turned into a SPIN model checker, which is a differenet thing.

Adam

> On Apr 15, 2026, at 6:40 AM, Marc Rochkind via TUHS <tuhs at tuhs.org> wrote:
> 
> When you use the phrase "UNIX engineers," to whom are you referring? At the
> beginning and for years afterwards, UNIX was a research project. If there
> was engineering, it came along much later.
> 
> On Tue, Apr 14, 2026, 10:41 PM Adam Koszek via TUHS <tuhs at tuhs.org> wrote:
> 
>> Hello,
>> 
>> A friend told me that UNIX engineers went through great pains to test
>> software. I heard a story about bubble gum wrapper being stuck into IDE
>> disk cable to short some lines to inject errors, and see if the filesystem
>> would handle/survive this.
>> 
>> Were there any more examples of this?
>> 
>> After UNIX was already running many program, were any of the complex
>> pieces of system code ever developed in the user-space first?
>> Filesystems/schedulers/networking are inherently hard to get right, and the
>> compile -> boot -> see kernel fail wasn’t ever fun cycle for me, and it had
>> to be hell on slow machines..
>> 
>> How was all this good code delivered? :)
>> 
>> I’ve learnt a buzzword: “deterministic simulation testing”. It’s when you
>> stub everything and run program in a fully controllable virtual world, and
>> can inject faults etc. These days I can build mini datacenter in a single
>> Golang program, but I’m wondering how the software that came before was
>> developed.
>> 
>> Adam