3.3. Section 3: Declarations and Types

Declarations and types are supported as specified in the Ada standard (See RM Section 3.9).

3.3.1. Uninitialized Variables

When the compile time options -Wuninitialized -O are used, the compiler will flag variables that may be uninitialized.

3.3.2. Enumeration Types

Enumeration types are supported as defined in the Ada 95 Reference Manual. Additional code and read-only data are generated to support the attributes 'Image, 'Pos and 'Val.

The size of enumeration objects is the minimum required to accommodate all the values, and including any representations given in a representation clause. The compiler will select a size of 8, 16 or 32 bits as appropriate.

Enumeration types may be packed to reduce wasted space in arrays of enumeration objects.

3.3.3. Integer Types

ERC32 Ada provides five predefined Integer types:

Table 3-1 gives the values of the attributes Size, First and Last for these types.

Table 3-1. Attributes of the Predefined Integer Types

Type nameSizeFirstLast
Short_Short_Integer8-2727-1
Short_Integer16-215215-1
Integer32-231231-1
Long_Integer32-231231-1
Long_Long_Integer64-263263-1

User-Defined Types. For a user-defined integer type, the compiler automatically selects the smallest compatible predefined integer type as the base type. For example, given the following type definition:


type My_Integer is range -10 .. +10;

the compiler will use Short_Short_Integer as the base type, and My_Integer'Size will be 8 bits.

Modular Types. ERC32 Ada supports modular types up to 64 bits in size. Like the integer types, these are represented in 8, 16, 32 or 64 bits as appropriate. The following declarations are legal:


type word_8 is mod 256;
type word_16 is mod 65536;
type word_32 is mod 2**32;
type word_64 is mod 2**64;

The standard Ada 95 operators for modular types are supported.

3.3.4. Floating Point Types

ERC32 Ada provides four predefined floating-point types:

The types Short_Float and Float are represented by the 32-bit single precision IEEE format; the types Long_Float and Long_Long_Float are represented by the 64-bit IEEE format. Note that the IEEE 80-bit extended precision format is not supported by the ERC32.

Table 3-2 gives the values of the attributes for the predefined floating-point types.

Table 3-2. Basic Attributes of Floating Point Types

AttributeFloatLong_Float
Size3264
Digits615
Machine_Radix22
Machine_Mantissa2352
Machine_Emax1281024
Machine_Emin-125-1021
Machine_RoundsFalseFalse
Machine_OverflowsFalseFalse

3.3.5. Fixed Point Types

ERC32 Ada supports fixed-point types up to 64 bits in size using 8, 16, 32 or 64 bits as appropriate. The value of 'Small may be either a power of two, or an arbitrary value given in a representation clause.