qcheck-core

Core qcheck library
Library qcheck-core
Module QCheck2 . Shrink
module type Number = sig ... end

Util module representing a number type, used for ad hoc polymorphism of some functions like number_towards.

val number_towards : (module Number with type t = 'a) -> destination:'a -> 'a -> 'a Seq.t

Shrink a number by edging towards a destination.

The destination is always the first value for optimal shrinking.

let int64_towards_list destination x = List.of_seq @@
  Gen.number_towards (module Int64) ~destination x
in
assert (int64_towards_list 0L 100L =
        [0L; 50L; 75L; 88L; 94L; 97L; 99L]);
assert (int64_towards_list 500L 1000L =
        [500L; 750L; 875L; 938L; 969L; 985L; 993L; 997L; 999L]);
assert (int64_towards_list (-50L) (-26L) =
        [-50L; -38L; -32L; -29L; -28L; -27L])

This generic function is exposed to let users reuse this shrinking technique for their custom number types. More specialized, convenient functions are provided below, e.g. int_towards.

val int_towards : int -> int -> int Seq.t

number_towards specialized to int.

val int32_towards : int32 -> int32 -> int32 Seq.t

number_towards specialized to int32.

val int64_towards : int64 -> int64 -> int64 Seq.t

number_towards specialized to int64.

val float_towards : float -> float -> float Seq.t

number_towards specialized to float.

There are various ways to shrink a float:

  • try removing floating digits, i.e. towards integer values
  • try to get as close as possible to the destination, no matter the number of digits
  • a mix of both

This implementation, as it relies on the generic number_towards function, tries to get as close as possible to the destination, e.g. the last value of Gen.float_towards 50 100 may be 99.9969482421875 (or a similar value).

val int_aggressive_towards : int -> int -> int Seq.t

int_agressive_towards destination n gives all integers from destination to n (excluded).

Be careful about time and memory as the resulting list can be huge

val int_aggressive : int -> int Seq.t
  • deprecated

    Use int_aggressive_towards 0 instead.

  • since 0.7