The SONAME changed upstream and apparently this indicates ABI breakage even if the minor version changes there. See: https://abi-laboratory.pro/index.php?view=compat_report&l=xxhash&v1=0.7.4&v2=0.8.0&obj=8e24f&kind=abi It looks like we should bump the subslot and (ideally then) add subslot operators to consumers.
The SONAME change is a false positive, should still be libxxhash.so.0, but the change in XXH3_state_s likely warrants the suggested change to bump subslot.
(In reply to Guilherme Amadio from comment #1) > The SONAME change is a false positive, should still be libxxhash.so.0, but > the change in XXH3_state_s likely warrants the suggested change to bump > subslot. Let's bump it next time around to ensure people get rebuilds then, for safety.
As far as I can tell, the subslot bump was never needed. It was just a false positive as the ABI didn't actually change.