Syntax :

• Numbers : using int and float of 32 bits.
  Syntax :

  

• Vector, matrices or tensors : using arrays of objects. The limitation to mono-dimensional arrays is guided by the following reasons :
  • Access to multi-dimensional arrays implicitly requires multiplications for the index computation. However, most of the time, the present index can be computed from its previous value using a simple addition, depending on the algorithm.

    As a consequence the computation of these indexes must not be specified here, but at a higher level.

  • Several matrices such as symmetric matrices, diagonal matrices, triangular matrices have a special storage with a particular access function. Again this is not to be defined at this level.

  Furthermore, the index of an array -at this level of specification- can only be a constant (i.e. you can write T[24/2] or T[1+3*5]), or a bounded variable (i.e. to accept to compiling T[i] for an array T[SIZE], the system must know that ).

  Syntax :

  

• Data structures : using struct, including bit-fields, and union of objects. This allows to entirely define the structure of a portion of memory, and the addresses at which each component can be founded.

  Syntax :

  

• Memory-objects : objects builded from these elements are called memory objects, they represent the structure of a part of the processor memory or the structure of a message.
  :