2016-09-08 21:13:10 +04:00
|
|
|
(**************************************************************************)
|
|
|
|
(* *)
|
|
|
|
(* OCaml *)
|
|
|
|
(* *)
|
|
|
|
(* Xavier Leroy, projet Cristal, INRIA Rocquencourt *)
|
|
|
|
(* *)
|
|
|
|
(* Copyright 1996 Institut National de Recherche en Informatique et *)
|
|
|
|
(* en Automatique. *)
|
|
|
|
(* *)
|
|
|
|
(* All rights reserved. This file is distributed under the terms of *)
|
|
|
|
(* the GNU Lesser General Public License version 2.1, with the *)
|
|
|
|
(* special exception on linking described in the file LICENSE. *)
|
|
|
|
(* *)
|
|
|
|
(**************************************************************************)
|
|
|
|
|
2017-03-17 17:39:31 +04:00
|
|
|
(* TEZOS CHANGES
|
|
|
|
|
|
|
|
* Import version 4.04.0
|
|
|
|
* Remove deprecated functions
|
|
|
|
|
|
|
|
*)
|
2016-09-08 21:13:10 +04:00
|
|
|
(** 64-bit integers.
|
|
|
|
|
2017-11-13 19:34:00 +04:00
|
|
|
This module provides operations on the type [int64] of
|
|
|
|
signed 64-bit integers. Unlike the built-in [int] type,
|
|
|
|
the type [int64] is guaranteed to be exactly 64-bit wide on all
|
|
|
|
platforms. All arithmetic operations over [int64] are taken
|
|
|
|
modulo 2{^64}
|
|
|
|
|
|
|
|
Performance notice: values of type [int64] occupy more memory
|
|
|
|
space than values of type [int], and arithmetic operations on
|
|
|
|
[int64] are generally slower than those on [int]. Use [int64]
|
|
|
|
only when the application requires exact 64-bit arithmetic.
|
2016-09-08 21:13:10 +04:00
|
|
|
*)
|
|
|
|
|
|
|
|
val zero : int64
|
|
|
|
(** The 64-bit integer 0. *)
|
|
|
|
|
|
|
|
val one : int64
|
|
|
|
(** The 64-bit integer 1. *)
|
|
|
|
|
|
|
|
val minus_one : int64
|
|
|
|
(** The 64-bit integer -1. *)
|
|
|
|
|
|
|
|
external neg : int64 -> int64 = "%int64_neg"
|
|
|
|
(** Unary negation. *)
|
|
|
|
|
|
|
|
external add : int64 -> int64 -> int64 = "%int64_add"
|
|
|
|
(** Addition. *)
|
|
|
|
|
|
|
|
external sub : int64 -> int64 -> int64 = "%int64_sub"
|
|
|
|
(** Subtraction. *)
|
|
|
|
|
|
|
|
external mul : int64 -> int64 -> int64 = "%int64_mul"
|
|
|
|
(** Multiplication. *)
|
|
|
|
|
|
|
|
external div : int64 -> int64 -> int64 = "%int64_div"
|
|
|
|
(** Integer division. Raise [Division_by_zero] if the second
|
2017-11-13 19:34:00 +04:00
|
|
|
argument is zero. This division rounds the real quotient of
|
|
|
|
its arguments towards zero, as specified for {!Pervasives.(/)}. *)
|
2016-09-08 21:13:10 +04:00
|
|
|
|
|
|
|
external rem : int64 -> int64 -> int64 = "%int64_mod"
|
|
|
|
(** Integer remainder. If [y] is not zero, the result
|
2017-11-13 19:34:00 +04:00
|
|
|
of [Int64.rem x y] satisfies the following property:
|
|
|
|
[x = Int64.add (Int64.mul (Int64.div x y) y) (Int64.rem x y)].
|
|
|
|
If [y = 0], [Int64.rem x y] raises [Division_by_zero]. *)
|
2016-09-08 21:13:10 +04:00
|
|
|
|
|
|
|
val succ : int64 -> int64
|
|
|
|
(** Successor. [Int64.succ x] is [Int64.add x Int64.one]. *)
|
|
|
|
|
|
|
|
val pred : int64 -> int64
|
|
|
|
(** Predecessor. [Int64.pred x] is [Int64.sub x Int64.one]. *)
|
|
|
|
|
|
|
|
val abs : int64 -> int64
|
|
|
|
(** Return the absolute value of its argument. *)
|
|
|
|
|
|
|
|
val max_int : int64
|
|
|
|
(** The greatest representable 64-bit integer, 2{^63} - 1. *)
|
|
|
|
|
|
|
|
val min_int : int64
|
|
|
|
(** The smallest representable 64-bit integer, -2{^63}. *)
|
|
|
|
|
|
|
|
external logand : int64 -> int64 -> int64 = "%int64_and"
|
|
|
|
(** Bitwise logical and. *)
|
|
|
|
|
|
|
|
external logor : int64 -> int64 -> int64 = "%int64_or"
|
|
|
|
(** Bitwise logical or. *)
|
|
|
|
|
|
|
|
external logxor : int64 -> int64 -> int64 = "%int64_xor"
|
|
|
|
(** Bitwise logical exclusive or. *)
|
|
|
|
|
|
|
|
val lognot : int64 -> int64
|
|
|
|
(** Bitwise logical negation *)
|
|
|
|
|
|
|
|
external shift_left : int64 -> int -> int64 = "%int64_lsl"
|
|
|
|
(** [Int64.shift_left x y] shifts [x] to the left by [y] bits.
|
2017-11-13 19:34:00 +04:00
|
|
|
The result is unspecified if [y < 0] or [y >= 64]. *)
|
2016-09-08 21:13:10 +04:00
|
|
|
|
|
|
|
external shift_right : int64 -> int -> int64 = "%int64_asr"
|
|
|
|
(** [Int64.shift_right x y] shifts [x] to the right by [y] bits.
|
2017-11-13 19:34:00 +04:00
|
|
|
This is an arithmetic shift: the sign bit of [x] is replicated
|
|
|
|
and inserted in the vacated bits.
|
|
|
|
The result is unspecified if [y < 0] or [y >= 64]. *)
|
2016-09-08 21:13:10 +04:00
|
|
|
|
|
|
|
external shift_right_logical : int64 -> int -> int64 = "%int64_lsr"
|
|
|
|
(** [Int64.shift_right_logical x y] shifts [x] to the right by [y] bits.
|
2017-11-13 19:34:00 +04:00
|
|
|
This is a logical shift: zeroes are inserted in the vacated bits
|
|
|
|
regardless of the sign of [x].
|
|
|
|
The result is unspecified if [y < 0] or [y >= 64]. *)
|
2016-09-08 21:13:10 +04:00
|
|
|
|
|
|
|
external of_int : int -> int64 = "%int64_of_int"
|
|
|
|
(** Convert the given integer (type [int]) to a 64-bit integer
|
|
|
|
(type [int64]). *)
|
|
|
|
|
|
|
|
external to_int : int64 -> int = "%int64_to_int"
|
|
|
|
(** Convert the given 64-bit integer (type [int64]) to an
|
2017-11-13 19:34:00 +04:00
|
|
|
integer (type [int]). On 64-bit platforms, the 64-bit integer
|
|
|
|
is taken modulo 2{^63}, i.e. the high-order bit is lost
|
|
|
|
during the conversion. On 32-bit platforms, the 64-bit integer
|
|
|
|
is taken modulo 2{^31}, i.e. the top 33 bits are lost
|
|
|
|
during the conversion. *)
|
2016-09-08 21:13:10 +04:00
|
|
|
|
|
|
|
external of_float : float -> int64
|
|
|
|
= "caml_int64_of_float" "caml_int64_of_float_unboxed"
|
2017-11-13 19:34:00 +04:00
|
|
|
[@@unboxed] [@@noalloc]
|
2016-09-08 21:13:10 +04:00
|
|
|
(** Convert the given floating-point number to a 64-bit integer,
|
2017-11-13 19:34:00 +04:00
|
|
|
discarding the fractional part (truncate towards 0).
|
|
|
|
The result of the conversion is undefined if, after truncation,
|
|
|
|
the number is outside the range \[{!Int64.min_int}, {!Int64.max_int}\]. *)
|
2016-09-08 21:13:10 +04:00
|
|
|
|
|
|
|
external to_float : int64 -> float
|
|
|
|
= "caml_int64_to_float" "caml_int64_to_float_unboxed"
|
2017-11-13 19:34:00 +04:00
|
|
|
[@@unboxed] [@@noalloc]
|
2016-09-08 21:13:10 +04:00
|
|
|
(** Convert the given 64-bit integer to a floating-point number. *)
|
|
|
|
|
|
|
|
|
|
|
|
external of_int32 : int32 -> int64 = "%int64_of_int32"
|
|
|
|
(** Convert the given 32-bit integer (type [int32])
|
2017-11-13 19:34:00 +04:00
|
|
|
to a 64-bit integer (type [int64]). *)
|
2016-09-08 21:13:10 +04:00
|
|
|
|
|
|
|
external to_int32 : int64 -> int32 = "%int64_to_int32"
|
|
|
|
(** Convert the given 64-bit integer (type [int64]) to a
|
2017-11-13 19:34:00 +04:00
|
|
|
32-bit integer (type [int32]). The 64-bit integer
|
|
|
|
is taken modulo 2{^32}, i.e. the top 32 bits are lost
|
|
|
|
during the conversion. *)
|
2016-09-08 21:13:10 +04:00
|
|
|
|
|
|
|
external of_nativeint : nativeint -> int64 = "%int64_of_nativeint"
|
|
|
|
(** Convert the given native integer (type [nativeint])
|
2017-11-13 19:34:00 +04:00
|
|
|
to a 64-bit integer (type [int64]). *)
|
2016-09-08 21:13:10 +04:00
|
|
|
|
|
|
|
external to_nativeint : int64 -> nativeint = "%int64_to_nativeint"
|
|
|
|
(** Convert the given 64-bit integer (type [int64]) to a
|
2017-11-13 19:34:00 +04:00
|
|
|
native integer. On 32-bit platforms, the 64-bit integer
|
|
|
|
is taken modulo 2{^32}. On 64-bit platforms,
|
|
|
|
the conversion is exact. *)
|
2016-09-08 21:13:10 +04:00
|
|
|
|
|
|
|
external of_string : string -> int64 = "caml_int64_of_string"
|
|
|
|
(** Convert the given string to a 64-bit integer.
|
2017-11-13 19:34:00 +04:00
|
|
|
The string is read in decimal (by default) or in hexadecimal,
|
|
|
|
octal or binary if the string begins with [0x], [0o] or [0b]
|
|
|
|
respectively.
|
|
|
|
Raise [Failure "int_of_string"] if the given string is not
|
|
|
|
a valid representation of an integer, or if the integer represented
|
|
|
|
exceeds the range of integers representable in type [int64]. *)
|
2016-09-08 21:13:10 +04:00
|
|
|
|
|
|
|
val to_string : int64 -> string
|
|
|
|
(** Return the string representation of its argument, in decimal. *)
|
|
|
|
|
|
|
|
external bits_of_float : float -> int64
|
|
|
|
= "caml_int64_bits_of_float" "caml_int64_bits_of_float_unboxed"
|
2017-11-13 19:34:00 +04:00
|
|
|
[@@unboxed] [@@noalloc]
|
2016-09-08 21:13:10 +04:00
|
|
|
(** Return the internal representation of the given float according
|
2017-11-13 19:34:00 +04:00
|
|
|
to the IEEE 754 floating-point 'double format' bit layout.
|
|
|
|
Bit 63 of the result represents the sign of the float;
|
|
|
|
bits 62 to 52 represent the (biased) exponent; bits 51 to 0
|
|
|
|
represent the mantissa. *)
|
2016-09-08 21:13:10 +04:00
|
|
|
|
|
|
|
external float_of_bits : int64 -> float
|
|
|
|
= "caml_int64_float_of_bits" "caml_int64_float_of_bits_unboxed"
|
2017-11-13 19:34:00 +04:00
|
|
|
[@@unboxed] [@@noalloc]
|
2016-09-08 21:13:10 +04:00
|
|
|
(** Return the floating-point number whose internal representation,
|
2017-11-13 19:34:00 +04:00
|
|
|
according to the IEEE 754 floating-point 'double format' bit layout,
|
|
|
|
is the given [int64]. *)
|
2016-09-08 21:13:10 +04:00
|
|
|
|
|
|
|
type t = int64
|
|
|
|
(** An alias for the type of 64-bit integers. *)
|
|
|
|
|
|
|
|
val compare: t -> t -> int
|
|
|
|
(** The comparison function for 64-bit integers, with the same specification as
|
|
|
|
{!Pervasives.compare}. Along with the type [t], this function [compare]
|
|
|
|
allows the module [Int64] to be passed as argument to the functors
|
|
|
|
{!Set.Make} and {!Map.Make}. *)
|
|
|
|
|
|
|
|
val equal: t -> t -> bool
|
|
|
|
(** The equal function for int64s.
|
|
|
|
@since 4.03.0 *)
|