Verifying distributed systems with Stateright
ToolingComments
This sounds incredible... but does the state space explosion make it impossible for larger systems? I wonder where the actual limit is before the model checker just hangs...
The visual explorer is the real win here. Reading logs to find a race condition is prehistoric. Why bother with text when you can map the state?
Regarding the state space exploration, does it support checking for liveness properties, or is it limited to safety invariants? I am interested in how it manages fairness constraints during actor scheduling.
This feels like the 2010s push for TLA+ in every distributed project. Most of those models became stale documentation the second the first production hotfix was pushed.
It is an implementation-level checker. That solves the spec-to-code drift that killed previous verification efforts.