Hacker News Reader

Comment by bjourne | Hacker News Reader

Since this thread is already full of armchair airplane engineers opinions about the root cause, I'll add mine too: Never ever write new software under the condition that it must work EXACTLY as old software. It can't be done and you'll always miss something.

Case in point: Wine. They've been at it for years and the emulation layer is still far from perfect. It works 99% of the time which is good enough for games, but I wager not for airplane control systems.

View on HN

Wine is not a good example. They don't have access to the source code of the original system, and they haven't had the resources to cover the entire footprint of the Windows APIs.

That's a good point. It'd be like comparing Boeing building a new plane to, say, Tupolev trying to build an identical copy without any blueprints.

> Tupolev trying to build an identical copy without any blueprints.

They actually did. It's called Tu-4: https://en.wikipedia.org/wiki/Tupolev_Tu-4

Fair point. Compaq in 1980's is probably the best case for reverse engineering and building a robust system. In some cases it out performed IBM PC's.

However, I have to admit, reverse engineering a home desktop is different from a airline software, considering that bugs can instantly kill people.

Also, if you rebuild a system in isolation, you have to reproduce all the bugs - since you do not know which of the bugs are load-bearing, i.e. other systems have been engineered to depend on the bugs, whether by accident or intentionally.

Absolutely correct. (cough) javascript at Microsoft.

GNU Octave would like to disagree.

In fact, divergence from MATLAB is considered a bug. Admittedly, the entire point there is a free and open source replication of MATLAB, and is only really necessitated by licensing.

No, the Octave faq reads: "There are still a number of differences between Octave and Matlab, however in general differences between the two are considered as bugs. Octave might consider that the bug is in Matlab and do nothing about it, but generally functionality is almost identical." Octave doesn't (for good reason!) re-implement bugs found in Matlab meaning it doesn't and never will work exactly like Matlab.

Boeing's engineers weren't tasked with making MCAS generally work like a 737. They were tasked with making MCAS work exactly like a 30+ year old aircraft model, complete with quirks and faults. An impossible task.

Alright, I think I see the point you're going for now. I think I misinterpreted what you were saying before.

You're saying they set out to make not-a-737 behave physically as a 737. Yes. I concur. I seem to have hyperfocusef on the replication part and not enough on the plane part.

Reply on news.ycombinator.com