Hacker News new | ask | show | jobs
by machine_coffee 737 days ago
Something crashes, it's _never_ a compiler bug ...

April 16 Oh, it actually is a compiler bug!

Can confidently say I've never experienced one in my career, but at some level of abstraction I guess it's less rare to find them.
2 comments
anyfoo 737 days ago
It's never a compiler bug, and it's definitely never a CPU bug.

Unless you're a low level kernel developer, then it's both!

(Jokes aside, it's still much, much more often not than it is, but it definitely does happen.)

link
mschuster91 737 days ago
> Unless you're a low level kernel developer, then it's both!

Or you're working in cryptography code and have to fight against compilers optimizing away sensitive stuff [1] - or adding junk back in [2].

[1] https://www.redhat.com/en/blog/security-flaws-caused-compile...

[2] https://pqshield.com/pqshield-plugs-timing-leaks-in-kyber-ml...

link
mort96 737 days ago
I mean she wrote the compiler didn't she? At least the register allocator part

"It's never a compiler bug" doesn't really apply when it's your own work-in-progress compiler you're testing...

link