←back to thread

War story: the hardest bug I ever debugged

(www.clientserver.dev)
462 points jakevoytko | 6 comments | 24 Mar 25 14:37 UTC | HN request time: 0.745s | source | bottom
Show context
BobbyTables2 ◴[27 Mar 25 03:27 UTC] No.43490119[source]▶
Interesting writeup, but 2 days to debug “the hardest bug ever”, while accurate, seems a bit overdone.

Though abs() returning negative numbers is hilarious.. “You had one job…”

To me, the hardest bugs are nearly irreproducible “Heisenbugs” that vanish when instrumentation is added.

I’m not just talking about concurrency issues either…

The kind of bug where a reproduction attempt takes a week, not parallelizable due to HW constraints, and logging instrumentation makes it go away or fail differently.

2 days is cute though.

replies(13): >>43490149 #>>43490287 #>>43490459 #>>43490557 #>>43491079 #>>43491823 #>>43492539 #>>43492555 #>>43492647 #>>43493115 #>>43493245 #>>43493811 #>>43497018 #
1. jffhn ◴[27 Mar 25 07:07 UTC] No.43491079[source]▶
>Though abs() returning negative numbers is hilarious.

Math.abs(Integer.MIN_VALUE) in Java very seriously returns -2147483648, as there is no int for 2147483648.

replies(5): >>43491454 #>>43491920 #>>43494203 #>>43497622 #>>43499425 #
2. eterm ◴[27 Mar 25 08:31 UTC] No.43491454[source]▶
You inspired me to check what .NET does in that situation.

It throws an OverflowException: ("Negating the minimum value of a twos complement number is invalid.")

3. adrian_b ◴[27 Mar 25 10:03 UTC] No.43491920[source]▶
Unchecked integer overflow strikes again.
4. rhaps0dy ◴[27 Mar 25 14:44 UTC] No.43494203[source]▶
Oh no, Pytorch does the same thing:

a = torch.tensor(-2*31, dtype=torch.int32) assert a == a.abs()

replies(1): >>43497253 #
5. MawKKe ◴[27 Mar 25 19:36 UTC] No.43497253[source]▶
numpy as well. and tensorflow
6. bobbylarrybobby ◴[27 Mar 25 23:24 UTC] No.43499425[source]▶
Rust does the same in release, although it panics in debug.