Foreword

This document describes Lucid Synchrone, a dedicated to the implementation of reactive systems. Lucid Synchrone ¹ is an ML-extension of the synchronous data-flow language Lustre [4]. The main features of the language are the following:

• The language has a data-flow flavor à la Lustre and the syntax is largely reminiscent to the one of Objective Caml [7]. It manages infinite sequences or streams as primitive values. These streams are used for representing input and output signals of a reactive system.
• The language provides some classical features of ML languages such as higher-order (a function can be parameterized by a function or return a function) or type inference.
• The language is build on the notion of clocks as a way to specify different execution rates in a program. In Lucid Synchrone, clocks are types and are computed automatically.
• Two program analysis are also provided. A causality analysis rejects programs which cannot be statically scheduled and an initialization analysis rejects programs whose behavior depends on uninitialized delays.
• The language allows for the definition of data-types: product types, record types and sum types. Structured values can be accessed through a pattern matching construction.
• Programs are compiled into Objective Caml. When required (through a compilation flag), the compiler ensures that the resulting program is “real-time”, i.e., it uses bounded memory and has a bounded response time for all possible program inputs.
• A module system is provided for importing values from the host language Objective Caml or from other synchronous modules.

Version 3 is a major revision and offers new features with respect to versions 1.0 and 2.0.

• The language allows to combine data-flow equations with complex state machines with various forms of transitions).
• Activation conditions are done through a pattern matching mechanism.
• Besides the regular delay primitive pre, a new delay primitive called last has been added in order to make the communication between shared variables in control structures (activation conditions or automata) easier.
• The language provides a way to define static values (i.e., infinite constant sequences). These static values may be arbitrarily complex but the compiler guaranty that they can be computed once for all, at instantiation time, before the first reaction starts.
• It is possible to define valued signals. Signals are stream values paired with a presence information (called enable in circuit terminology). The value of signal can be accessed through a pattern matching construct.
• The language supports streams of functions as a way to describe reconfigurable systems.
• Sequential functions (as opposed to combinatorial function to keep circuit terminology) must now be explicitly declared with the keyword node. Otherwise, they are considered to be combinatorial.
• Internally, the compiler has been completely rewritten. We abandoned the compilation method introduced in version 2.0 and came back to the (co-iteration based) compilation method introduced in version 1.0.

Availability

The current implementation is written in Objective Caml. The use of the language needs the installation of Objective Caml.

The language is experimental and evolves continuously. Please send comments or bug to Marc.Pouzet@lri.fr.

Copyright notice

This software includes the Objective Caml run-time system, which is copyrighted INRIA, 2006.