Software bugs causing deaths

>>59524583
an ariane 5 rocket crashed because of an integer overflow
http://sunnyday.mit.edu/accidents/Ariane5accidentreport.html

>>59524694
Topkek, only after 40 seconds.
Plebs.

Did software failure play any part in Chernobyl, Fukushima, or Three Mile Island or were those all mechanical failures?

>>59524806
Chernobyl was basically doing a stress test of the reactor while having ALL safety devices turned off. No software failure or mechanical failure at all really.

>>59524806
Chernobyl was pure bumblefuckery. The facility was horribly maintained and the workers had no fucking clue what they were doing. From what I've read, the plant itself wasn't poorly designed, but when you deliberately bypass each and every safety measure put in place to prevent retards from nuking themselves entirely because they were working as designed shit's bound to blow up.

>>59524583

https://arstechnica.com/information-technology/2015/05/boeing-787-dreamliners-contain-a-potentially-catastrophic-software-bug/

>>59524694

Nobody died. Not a big deal.

>fMRI bug went undetected/ignored for FIFTEEN FUCKIN YEARS

http://www.sciencealert.com/a-bug-in-fmri-software-could-invalidate-decades-of-brain-research-scientists-discover

>>59524806
Chernobyl happened when the operators turned the safeties off because they were planning on pressing the 'blow up reactor' button and didn't want the safeties to ruin the experiment.

Fukushima happened when a nuclear reactor (pre chernobyl, that thing was old) was hit by an earthquake and a tsunami at the same time, and the backup generators (for cooling) were flooded.

Three mile island... I can't actually remember but it was user error down to the bone. I think it was something to do with someone slamming off the safeties because 'they knew better'

Deaths? Eh. Not really. There was some article about a massive amount of psychology MRI data being fucking useless because the algorithm in some analysis software wasn't written properly and people were using it for many years.

The Patriot missile system has confirmed deaths from leaving the machine running too long and letting a known floating point rounding error accumulate to the point it couldn't intercept enemy mortars.

The genius programers thought scanning radar every .1 seconds would be good, but didn't realize that .1 is irrational in base 2, and every .1 seconds the timing would be slightly off

>>59524999
I remember read a different article that suggested people may have been misdiagnosed with shit like schizophrenia and epilepsy based on MRI scans. They said something about brain cancer as well. Im too lazy to track it down though. Also i dont know shit about modern medicine so I don't know if the above is even true.

>>59525061

fucking jews

>>59524974
still billions down the drain because someone thought it wouldn't be a big deal to cast 64b float to 16b int

File: 3dClustSim.png (227KB, 1177x2100px) Image search: [Google]

227KB, 1177x2100px

>>59525066
>>59524999
Pic related is some of the code of the program in question, including some actual comments from it.

Though actually the bugs were a minor factor in the data which was rendered useless; the big issue were some incorrect statistical assumptions which led to a vast underestimation of the error rate.

>>59524583
>What are some of the most severe results to occur from software bugs/malfunctioning?
uint16_t overflow caused lumbermill to kill a guy once.

>>59524806
Chernobyl blew up because of human error combined with incompetence.

Fukushima happened because Japanese culture prevented workers from letting decision makes know that something was about to go really bad.

>>59525448
>C fizzbuzzers will unironically call this good code

>>59525448
Also we're talking fMRI specifically so we're talking things like research of what part of the brain is involved in what processes etc. There were basically many more false positives than was initially believed and since data archiving practices in the field were bad for most results there is no way to go back and re-evaluate them.

Babby's intro to software engineering homework

>>59524982

Three Mile Island, if I recall correctly, happened because somebody decided its good idea to save absurdly low ammount of money by skipping things like sensor informing, if pressure control systems works properly. One day it didnt, while operator had no idea what happened, except that he knew about pressure dropping inside reactor. First guess was that reaction itself is "dying", so he started to turn it up. When they realized it was just simple mechanical error, it was too late.

Or Im recalling completely different incident.

>>59524982
>Turned the safeties off because they were planning on pressing the 'blow up reactor' button and didn't want the safeties to ruin the experiment.

So kind of like writing anything critical in C?

>>59524694
IIRC, what crashed it wasn't even the integer overflow itself (since that part of the code was actually non-critical), but the uncaught exception caused by it.

Don't have your functions throwing exceptions in your release versions guys. Make sure your software can recover from a failure.

>>59525613
>Babby's
Baby's
>software engineering
programming

And stop using a tripcode. No one cares who you are.

>>59525716
>No one cares who you are.
until you put on the mask.

>>59525061
>The genius programers thought scanning radar every .1 seconds would be good, but didn't realize that .1 is irrational in base 2, and every .1 seconds the timing would be slightly off
This would've been okay because the error would cancel but part of the code was using a newer more accurate conversion so no more cancelling.

>>59525739
For you

>>59525716
No, this is a subject in most entry level software engineering courses.

I guess you'd know that if you had to take one of these courses during your CS curriculum.

>>59525716
I care.

>>59525613
>implying
I'm not even in college atm, I was just reading a Wikipedia page about UNIX and it mentioned catastrophic software bugs and I thought it was interesting. Why am I replying to a tripfag?

>tfw tripfags ruin another thread

File: worry.jpg (17KB, 190x245px) Image search: [Google]

17KB, 190x245px

>>59525448
>that last comment

>>59527850
inb4 they reply to this with some retarded justification as to why they want to trip on a FUCKING anonymous image board.

>>59524583
https://en.wikipedia.org/wiki/Therac-25

>>59528631
>The system noticed that something was wrong and halted the X-ray beam, but merely displayed the word "MALFUNCTION" followed by a number from 1 to 64. The user manual did not explain or even address the error codes, so the operator pressed the P key to override the warning and proceed anyway.
user malfunction

>>59525536
>uint16_t overflow caused lumbermill to kill a guy once.
story?

>>59524583
Didn't Tesla self-driving cars killed somebody?

I remember being about not recognizing a white truck crashing against it, sounds like a bug to me.

>>59524583
can't wait to break into heart transplants via wi-fi

A patriot missile crashed into a military base after an integer overflow bug.

This is why software/computer engineers exist. You have someone to put in jail when they negligently kill a bunch of people.

>>59528917
IIRC, the driver fell asleep and the car couldn't see a white truck on a brightly lit skyline. It sounds like less of a bug and more of a limit on what the system was capable of, as well as user error.

>>59529022
> the car couldn't see a white truck on a brightly lit skyline

It probably didn't see it but it could have sense it, I'm pretty sure all self-driving cars have a ton of ultra-sonic and laser proximity sensors, so if the car decided to trust his vision over his sensors then it was probably a bug or at least shitty code.