Antescofo Intermediate Representation

We describe an intermediate language designed as a medium-level internal representation of programs of the interactive music system Antescofo. This representation is independent both of the Antescofo source language and of the architecture of the execution platform. It is used in tasks such as verification of timings, model-based conformance testing, static control-flow analysis or simulation. This language is essentially a flat representation of Antescofo’s code, as a finite state machine extended with local and global variables, with delays and with concurrent threads creation. It features a small number of simple instructions which are either blocking (wait for external event, signal or duration) or not (variable assignment, message emission and control). Key-words: Interactive Music Systems, Compilation, Formal Models This work has been supported by the ANR project Inedit (ANR-12-CORD-009) http://inedit.ircam.fr. ∗ INRIA † Ircam, UMR STMS 9912 CNRS/UPMC ‡ DGA & INRIA Représentation Intermédiaire pour le langage d’Antescofo Résumé : Ce rapport décrit un langage intermédiaire conçu pour la représentation interne de programmes du système musical interactif Antescofo. Il est actuellement utilisé dans des tâches de vérification portant en particulier sur les durées, de test de conformité fondé sur modèles, d’analyse statique et de simulation. Mots-clés : Systèmes musicaux interactifs, compilation, modèles formels Antescofo Intermediate Representation 3 1 Intermediate Code: Syntax We describe in this section an abstract syntax for the intermediate code which will be the result of a front-end compilation of Antescofo’s programs. It is defined independently of Antescofo’s source language and of the architecture of the execution platform. We give in the description some examples corresponding to the compilation of programs in Antescofo language. 1.1 Values 1.1.1 Atomic Values We assume the same scalar values as in Antescofo, see [5, 2]: Booleans values true and false , the integers, the floats (double), the strings and one undefined value (which is not used in this document). We also assume compounds values for vectors and maps. Durations are a specific type of value. They can be expressed with different time units, corresponding to different clocks. For instance, the seconds is the time unit of the wall clock (physical time). In Antescofo, the most important time unit is beats, which refers to an inferred tempo. As explained in Section 2.1, we assume that time units are inter-convertible and hence we shall sometimes drop them in the expression of delays in the following. 1.1.2 Variables Let Xg and Xl be two disjoint infinite sets of respectively global variables and local variables. 1.1.3 Expressions The expressions are the same as in the Antescofo language [5]. Note that predicates operating on duration values will rely on multi clock services described in Section 2.1. An expression is called ground when it does not contain variables. 1.2 Symbols 1.2.1 Input Symbols We assume a given set of input symbols I = {i0, . . .}, called input events, representing some information expected from the external environment. The set I is assumed totally ordered by a function called next . We take for instance a set of Antescofo’s events (notes etc), as defined in Section 2 of [5], together with their positions in the score. For such an input symbol i at position n, next(i) is then defined as the event at position n + 1. A generalization of the total ordering on input events into a DFA with state set Q and input alphabet I will be the subject of further work. 1.2.2 Output Symbols We assume a given set of symbols O = {a0, a1 . . .}, representing action emitted or messages sent to the external environment. For Antescofo, the elements of O are called internal (atomic) actions and can be messages to MAX/MSP, OSC messages...