ved-runtime

Deterministic execution engine for the Ved control-plane programming language.

Overview

ved-runtime implements the core execution model of Ved programs.

It is responsible for:

• deterministic scheduling of transition slices
• persistent state evolution through journal + snapshot mechanisms
• goal-driven reconciliation loops
• effect intent isolation and replayable execution
• crash recovery and logical time progression

The runtime is designed to support long-running orchestration systems that must continuously stabilize distributed software environments.

Design Goals

• Predictable and reproducible system behaviour
• Structured isolation between execution domains
• Explicit convergence toward declared goals
• Failure-resilient state persistence
• Minimal trusted core suitable for future distributed evolution

Status

Early prototype.

Initial milestones achieved:

• Bytecode execution loop
• Append-only journal & snapshots
• Deterministic scheduler and logical clocks
• Basic domain state model
• Developer Experience tools (structured traces, basic compiler errors)

CLI Usage

The ved-cli provides commands to run and inspect Ved programs:

• Run a program: cargo run -p ved-cli -- run path/to/file.ved This compiles the file, runs the scheduler, generates snapshot states (.snapshot.json), and creates an execution trace file (.trace.json).

• View a trace: cargo run -p ved-cli -- view-trace path/to/file.trace.json Prints a formatted, human-readable view of the exact sequence of domain state changes and messages.

• Compile only: cargo run -p ved-cli -- compile path/to/file.ved Validates syntax and semantics without running the scheduler.

Repository Structure (planned)

• scheduler
• executor / virtual machine
• persistence engine
• effect runtime
• observability utilities

Contributing

Design discussions and development are evolving rapidly. Please open issues before large implementation changes.

License

Apache License 2.0