Lexum

Your system behaves
exactly as written.

Deterministic control for distributed systems. Replace imperative orchestration chaos with infrastructure governed by executable law.

Explore Development Experience
View on Github
DOMAIN: SCHED-01

Deterministic Scheduler

[CONVERGING]
Transitions are calculated fully before partial application. The runtime guarantees convergence prior to side-effect execution.
DOMAIN: STATE-00

Snapshot Manager

[QUIESCENT]
Persistent DAG state verified. Awaiting upstream signals or mailbox deltas to resume execution loop.

Imperative orchestration
fails dynamically.

Modern infrastructure relies on YAML applied sequentially by imperative runners. When state drifts, they react chaotically. Lexum replaces reactions with computation.

MODEL: IMPERATIVE / REACTIVE[DRIFT DETECTED]

State mutated out of band. Race conditions inevitable. Convergence acts as a coincidental side-effect of retry loops.

MODEL: DETERMINISTIC / LAW[CONVERGED]

System advances through synchronized transitions. Configuration drift impossible. Convergence guaranteed by the runtime mathematically.

[ RUNTIME SPECIFICATION ]

Governed distributed uncertainty. Replayable, safe, convergent.

ARCH-01

Deterministic Scheduler

Evaluates dependency graphs strictly. No parallel branches execute without guaranteed eventual quiescence. Enforces order.

↳Deterministic Scheduler↳Scheduler Design
ARCH-02

Goal-Specification Semantics

Instead of writing step-by-step creation logic, state goals are defined. The semantic engine diffs reality against the target.

↳Goal Specification Semantics↳Goals and Convergence
ARCH-03

Spatial Authority Model

Variables lack global scope. Each resource is owned by an explicitly bounded domain, eliminating concurrent unmanaged mutations.

↳Spatial Authority Model↳Authority Scope Hierarchy
ARCH-04

Persistent Snapshot Engine

State is never assumed; it is captured, cryptographically hashed, and strictly stored to support exact point-in-time replay.

↳Persistent Snapshot Engine↳State Persistence Strategy
ARCH-05

Mailbox / Actor Coordination

Distributed actions pass through strict queues. Multi-domain coordination resolves before execution progresses, never during.

↳Mailbox Model↳Asynchronous Messaging
ARCH-06

Failure Recovery Loop

When real-world assertions crash, the runtime halts, logs the DAG diff, and rolls back the logical plane to the last locked quiescence.

↳Failure Recovery Loop↳Crash Recovery Model
[ TRACE REPLAY ENGINE ]

Inspect the syntax. Replay the state.

Because execution is deterministic, state can be stepped forward and backward mathematically. Slide the execution trace to verify the exact causal chain of the `.lxm` runtime.

EXECUTION_TRACE_V1.4
STEP: 0/4 [ INIT ]
1domain Network {
2 goal: TCP_Port(80)
3}
4
5domain Application {
6 depends_on: Network
7 goal: ReplicaSet(3)
8}
9
10transition Resolve_Topology {
11 assert Network.quiescent
12 execute => apply Application.goal
13}
DOMAIN: Network[INACTIVE]
Goal: TCP_Port(80)
DOMAIN: Application[INACTIVE]
Goal: ReplicaSet(3)
TIME_SCRUB
The Lexum Thesis
Math,notmagic.Buildsystemsthatconvergepredictablywithoutthechaosofimperativescripting.
Read the Whitepaper

Frequently Asked Questions

Common questions about the Lexum language and runtime.

Lexum is a deterministic control-plane programming language designed to help engineers build reliable, long-running distributed systems. It enables developers to describe desired system behaviour using structured state models, goals, and bounded execution logic.

[ SYSTEM READY ]

Initiate.

LEXUM_CLI
$ curl -sSf https://lexum.org/install.sh | sh

Includes the `lxm` compiler, runtime engine, and local trace visualizer.