Text.Fixity.Internal

Associativity of an operator.

Fixity of an operator.

We allow unary operators to have associativity, because it's useful to distinguish between an expression like -(-x) (unary minus, left-associative) and ~~b (boolean NOT, right-associative).

Associativity of unary operators also matters when pretty-printing expressions like (-x) + y, which is pretty-printed as -x + y, assuming unary minus has the same fixity as + (and both the operators are left-associative). I.e. unary minus is handled just like the binary one: (0 - x) + y is pretty-printed as 0 - x + y.

Postfix operators are handled similarly. E.g. if ! is left-associative, then (x!)! is pretty-printed as x!! and if it's right-associative -- (x!)!.

The data type is parameterized, so that the user can choose precedence to be integer/fractional, bounded/unbounded, etc (we could also allows operators to be partially or totally ordered, but at the moment prec is required to implement Ord, i.e. it has to be totally ordered). By default we go with bounded fractional precedence, see the main Text.Fixity module.

Direction in which pretty-printing goes. For example in x + y x is pretty-printed to the left of + and y is pretty-printed to the right of +.

A context that an expression is being rendered in.

Enclose an a (using the provided function) if required or leave it as is. The need for enclosing is determined from an outer RenderContext and the inner fixity.