Hacker News Reader

Comment by jmount | Hacker News Reader

Hopefully I am not too naive, but I think aircraft safety redundancy remains above retail car standards. Also, in aircraft they "have time to solve some problems", versus freeway bumper cars.

View on HN

I also don't believe they install OTA updates while in flight.

Although if they did it would give a fantastic new meaning to "over the air" :-)

In service to the pun, there is a relatively famous demo of using erlang for embedded development where they show off hot code reloading of a drone's flight software while it's in flight.

https://www.youtube.com/watch?v=XQS9SECCp1I

More to the point, FAS regulations would absolutely forbid any such event. They probably mandate testing of the updates before returning to airplane to service.

Completely unlike the safety standards for cars.

Which has nothing to do with a bad update being installed during routing maintenance and only failing in flight. Or while driving.

I am assuming/hoping that testing from Boeing/Airbus is far more stringent than that of Jeep !

Boeing doesn't exactly have the best reputation.

Yeah

Also people say "oh what if fly-by-wire fails" well what if traditional hydraulic controls fail, which has happened plenty in the history of commercial aviation

Everything can and will fail at some point

No redundancy is redundancy enough in some %0.xx of cases. You can always reduce the number, but never make it 0

The reliability of software is so bad this is an absurd comparison.

This is a safety standards issue not a "software" issue. Standards for airplane software are very high

Most planes have been fly-by-wire for decades and aren't regularly falling out of the sky

Software is much easier to break accidentally than a valve.

I work for a medical device manufacturer, and software absolutely can be designed to be just as reliable as physical systems, but the development and testing process looks completely different than a developing a mobile app. Things slow WAY down: if you want to change one line of code, it'll take literally weeks before it makes it to a production environment because of all the testing, documentation, justification, and human approvals. I imagine flight safety systems are subject to a similar level of rigor.

Obligatory mention of SQLight going the DO-178B route.

https://corecursive.com/066-sqlite-with-richard-hipp/#testin...

"Richard Hipp: Getting that last 5% is really, really hard and it took about a year for me to get there, but once we got to that point, we stopped getting bug reports from Android.

"Richard Hipp: Yes, so we’ll do billions of tests."

> I imagine flight safety systems are subject to a similar level of rigor.

Not for cars tho. Cars can be expected to be as reliable as the average windows PC unless there are regulations made and accountability assigned

Not for software. Hence airplanes needing reboots at certain time frequencies before they bug out in weird ways.

Indeed, but read the link I posted above if you're interested in a fascinating case of failed redundancy.

Engaging version by the incredible Admiral Cloudberg:

https://admiralcloudberg.medium.com/fields-of-fortune-the-cr...

Reply on news.ycombinator.com