|
|
|
|
|
|
by mcl0vinit
88 days ago
|
|
|
A bit more context on the design: The main thing I wanted was a stronger execution model than scripts + retries. In Vilano Runtime, the kernel owns durable truth and coordination: waits, timers, retries, service inbox state, supervision, and replay state. Workers are disposable and rerun orchestration code from the top, resolving durable operations through the kernel. That split is what lets workflows and long-lived keyed services recover cleanly across failure and restart. |
|
|