7 ms·

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, bu

• anyfoo 2y 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.)
  • mschuster91 2y 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-compiler-optimizations https://www.redhat.com/en/blog/security-flaws-caused-compile... [2] https://pqshield.com/pqshield-plugs-timing-leaks-in-kyber-ml-kem-to-improve-pqc-implementation-maturity/ https://pqshield.com/pqshield-plugs-timing-leaks-in-kyber-ml...
• mort96 2y 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...