2.3 KiB
Integrated types
In ++C, there are several integrated types:
voidint8_t,int16_t, ...int256_tfl16_t,fl32_t,bigfl_t
Void type
The void type is for all intents and purposes an empty structures. All other empty structs are convertible from and to a void. There is also a literal null, that represents a value of type void. All types that want to implement a default value will implement an explicit cast from void. void is the return value of all procedures (functions with no return values). All void variables/arguments will be omitted.
Integral types
An integral type is an integer, made up of any amount of bytes. It will by default define all integral operators:
- Addition / subtraction
- Multiplication / division / modulo
- Negating
- Bitwise AND, OR, XOR and flipping
- Comparisons
Although the underlying language/architecture might not support a certain size of integers, the translator will guarantee that these types will work as expected.
Float types
A float type is a number with a floating fraction point. Although it is not required, it is almost guaranteed that the float types are an implementation of the IEEE float standard. Still, some platforms that don't support float maths will have to implement functions to do so, which leaves it up to the translator author to decide how the floats will work. Of course, the float type implements all algebraic operations for integers, including some float-specific:
- Floor, ceiling, round
- Square root
- Logarithm
- Exponentiation
- Sine, cosine, tangent
- Int to float conversion
- Float to int conversion
There are tree variants of the float: fl32_t, fl64_t and bigfl_t. fl32_t is the closest representation of the IEEE single float that the system can support. fl64_t is the same as single, but for double floats. bigfl_t is the biggest float that the system supports. Usually, fl64_t and bigfl_t are the same, but sometimes fl32_t, fl64_t and bigfl_t overlap. fl32_t and fl64_t are guaranteed to be 32-bit and 64-bit respectively. bigfl_t is used for systems that natively work with bigger floats (for example, the x86_64 architecture supports 80-bit floats).
How it works
These types don't actually exist, but rather are converted to special statements (native operators), that later will be transpiled to the underlying language's counterparts.