An (almost) catastrophic OpenZFS bug and the humans that made it (and Rust is here too) • Rob Norris
https://despairlabs.com/blog/posts/2025-07-10-an-openzfs-bug-and-the-humans-that-made-it/
An (almost) catastrophic OpenZFS bug and the humans that made it (and Rust is here too) • Rob Norris
https://despairlabs.com/blog/posts/2025-07-10-an-openzfs-bug-and-the-humans-that-made-it/
physical_size and allocated_size instead of psize and asize.
I agree with everything said about how this would have been caught in Rust (incl. @kornel 's reply), but getting rid of habits formed and based on constraints of the 1970ies would also help.
```
warning: value assigned to `psize` is never read
|
| psize <<= ashift;
| ^^^^^
|
= help: maybe it is overwritten before being read?
= note: `#[warn(unused_assignments)]` on by default
```
The seemingly trivial `#[allow(warning)]` annotations have a major impact: they allow the compiler to lint about uncertain problems by default, because false positives can be muted easily and precisely on case-by-case basis.
A space for Bonfire maintainers and contributors to communicate