assemblyscript/std/assembly.d.ts
dcodeIO 113925fa7e Add a mechanism to hook into the compilation process with custom transforms, see #20
A custom transform is a node module that exports hooks called by the compiler on specific occasions, then being able to modify the intermediate results. Starting with 'afterParse' to modify the AST before compilation is performed.
2018-05-26 13:13:39 +02:00

587 lines
28 KiB
TypeScript

/**
* Environment definitions for compiling AssemblyScript to WebAssembly using asc.
* @module std/assembly
*//***/
// Types
/** An 8-bit signed integer. */
declare type i8 = number;
/** A 16-bit signed integer. */
declare type i16 = number;
/** A 32-bit signed integer. */
declare type i32 = number;
/** A 64-bit signed integer. */
declare type i64 = number;
/** A 32-bit signed integer when targeting 32-bit WebAssembly or a 64-bit signed integer when targeting 64-bit WebAssembly. */
declare type isize = number;
/** An 8-bit unsigned integer. */
declare type u8 = number;
/** A 16-bit unsigned integer. */
declare type u16 = number;
/** A 32-bit unsigned integer. */
declare type u32 = number;
/** A 64-bit unsigned integer. */
declare type u64 = number;
/** A 32-bit unsigned integer when targeting 32-bit WebAssembly or a 64-bit unsigned integer when targeting 64-bit WebAssembly. */
declare type usize = number;
/** A 1-bit unsigned integer. */
declare type bool = boolean | number;
/** A 32-bit float. */
declare type f32 = number;
/** A 64-bit float. */
declare type f64 = number;
/** Converts any other numeric value to an 8-bit signed integer. */
declare function i8(value: i8 | i16 | i32 | i64 | isize | u8 | u16 | u32 | u64 | usize | bool | f32 | f64): i8;
declare namespace i8 {
/** Smallest representable value. */
export const MIN_VALUE: i8;
/** Largest representable value. */
export const MAX_VALUE: i8;
}
/** Converts any other numeric value to a 16-bit signed integer. */
declare function i16(value: i8 | i16 | i32 | i64 | isize | u8 | u16 | u32 | u64 | usize | bool | f32 | f64): i8;
declare namespace i16 {
/** Smallest representable value. */
export const MIN_VALUE: i16;
/** Largest representable value. */
export const MAX_VALUE: i16;
}
/** Converts any other numeric value to a 32-bit signed integer. */
declare function i32(value: i8 | i16 | i32 | i64 | isize | u8 | u16 | u32 | u64 | usize | bool | f32 | f64): i32;
declare namespace i32 {
/** Smallest representable value. */
export const MIN_VALUE: i32;
/** Largest representable value. */
export const MAX_VALUE: i32;
/** Loads an 8-bit signed integer from memory and returns it as a 32-bit integer. */
export function load8_s(offset: usize, constantOffset?: usize): i32;
/** Loads an 8-bit unsigned integer from memory and returns it as a 32-bit integer. */
export function load8_u(offset: usize, constantOffset?: usize): i32;
/** Loads a 16-bit signed integer from memory and returns it as a 32-bit integer. */
export function load16_s(offset: usize, constantOffset?: usize): i32;
/** Loads a 16-bit unsigned integer from memory and returns it as a 32-bit integer. */
export function load16_u(offset: usize, constantOffset?: usize): i32;
/** Loads a 32-bit integer from memory. */
export function load(offset: usize, constantOffset?: usize): i32;
/** Stores a 32-bit integer to memory as an 8-bit integer. */
export function store8(offset: usize, value: i32, constantOffset?: usize): void;
/** Stores a 32-bit integer to memory as a 16-bit integer. */
export function store16(offset: usize, value: i32, constantOffset?: usize): void;
/** Stores a 32-bit integer to memory. */
export function store(offset: usize, value: i32, constantOffset?: usize): void;
}
/** Converts any other numeric value to a 64-bit signed integer. */
declare function i64(value: i8 | i16 | i32 | i64 | isize | u8 | u16 | u32 | u64 | usize | bool | f32 | f64): i64;
declare namespace i64 {
/** Smallest representable value. */
export const MIN_VALUE: i64;
/** Largest representable value. */
export const MAX_VALUE: i64;
/** Loads an 8-bit signed integer from memory and returns it as a 64-bit signed integer. */
export function load8_s(offset: usize, constantOffset?: usize): i64;
/** Loads an 8-bit unsigned integer from memory and returns it as a 64-bit unsigned integer. */
export function load8_u(offset: usize, constantOffset?: usize): u64;
/** Loads a 16-bit signed integer from memory and returns it as a 64-bit signed integer. */
export function load16_s(offset: usize, constantOffset?: usize): i64;
/** Loads a 16-bit unsigned integer from memory and returns it as a 64-bit unsigned integer. */
export function load16_u(offset: usize, constantOffset?: usize): u64;
/** Loads a 32-bit signed integer from memory and returns it as a 64-bit signed integer. */
export function load32_s(offset: usize, constantOffset?: usize): i64;
/** Loads a 32-bit unsigned integer from memory and returns it as a 64-bit unsigned integer. */
export function load32_u(offset: usize, constantOffset?: usize): u64;
/** Loads a 64-bit unsigned integer from memory. */
export function load(offset: usize, constantOffset?: usize): i64;
/** Stores a 64-bit integer to memory as an 8-bit integer. */
export function store8(offset: usize, value: i64, constantOffset?: usize): void;
/** Stores a 64-bit integer to memory as a 16-bit integer. */
export function store16(offset: usize, value: i64, constantOffset?: usize): void;
/** Stores a 64-bit integer to memory as a 32-bit integer. */
export function store32(offset: usize, value: i64, constantOffset?: usize): void;
/** Stores a 64-bit integer to memory. */
export function store(offset: usize, value: i64, constantOffset?: usize): void;
}
/** Converts any other numeric value to a 32-bit (in WASM32) respectivel 64-bit (in WASM64) signed integer. */
declare var isize: i32 | i64;
/** Converts any other numeric value to an 8-bit unsigned integer. */
declare function u8(value: i8 | i16 | i32 | i64 | isize | u8 | u16 | u32 | u64 | usize | bool | f32 | f64): i8;
declare namespace u8 {
/** Smallest representable value. */
export const MIN_VALUE: u8;
/** Largest representable value. */
export const MAX_VALUE: u8;
}
/** Converts any other numeric value to a 16-bit unsigned integer. */
declare function u16(value: i8 | i16 | i32 | i64 | isize | u8 | u16 | u32 | u64 | usize | bool | f32 | f64): i8;
declare namespace u16 {
/** Smallest representable value. */
export const MIN_VALUE: u16;
/** Largest representable value. */
export const MAX_VALUE: u16;
}
/** Converts any other numeric value to a 32-bit unsigned integer. */
declare function u32(value: i8 | i16 | i32 | i64 | isize | u8 | u16 | u32 | u64 | usize | bool | f32 | f64): i32;
declare namespace u32 {
/** Smallest representable value. */
export const MIN_VALUE: u32;
/** Largest representable value. */
export const MAX_VALUE: u32;
}
/** Converts any other numeric value to a 64-bit unsigned integer. */
declare function u64(value: i8 | i16 | i32 | i64 | isize | u8 | u16 | u32 | u64 | usize | bool | f32 | f64): i64;
declare namespace u64 {
/** Smallest representable value. */
export const MIN_VALUE: u64;
/** Largest representable value. */
export const MAX_VALUE: u64;
}
/** Converts any other numeric value to a 32-bit (in WASM32) respectivel 64-bit (in WASM64) unsigned integer. */
declare var usize: u32 | u64;
/** Converts any other numeric value to a 1-bit unsigned integer. */
declare function bool(value: i8 | i16 | i32 | i64 | isize | u8 | u16 | u32 | u64 | usize | bool | f32 | f64): bool;
declare namespace bool {
/** Smallest representable value. */
export const MIN_VALUE: bool;
/** Largest representable value. */
export const MAX_VALUE: bool;
}
/** Converts any other numeric value to a 32-bit float. */
declare function f32(value: i8 | i16 | i32 | i64 | isize | u8 | u16 | u32 | u64 | usize | bool | f32 | f64): f32;
declare namespace f32 {
/** Smallest representable value. */
export const MIN_VALUE: f32;
/** Largest representable value. */
export const MAX_VALUE: f32;
/** Smallest normalized positive value. */
export const MIN_POSITIVE_VALUE: f32;
/** Smallest safely representable integer value. */
export const MIN_SAFE_INTEGER: f32;
/** Largest safely representable integer value. */
export const MAX_SAFE_INTEGER: f32;
/** Difference between 1 and the smallest representable value greater than 1. */
export const EPSILON: f32;
/** Returns the floating-point remainder of `x / y` (rounded towards zero). */
export function mod(x: f32, y: f32): f32;
/** Returns the floating-point remainder of `x / y` (rounded to nearest). */
export function rem(x: f32, y: f32): f32;
/** Loads a 32-bit float from memory. */
export function load(offset: usize, constantOffset?: usize): f32;
/** Stores a 32-bit float to memory. */
export function store(offset: usize, value: f32, constantOffset?: usize): void;
}
/** Converts any other numeric value to a 64-bit float. */
declare function f64(value: i8 | i16 | i32 | i64 | isize | u8 | u16 | u32 | u64 | usize | bool | f32 | f64): f64;
declare namespace f64 {
/** Smallest representable value. */
export const MIN_VALUE: f64;
/** Largest representable value. */
export const MAX_VALUE: f64;
/** Smallest normalized positive value. */
export const MIN_POSITIVE_VALUE: f64;
/** Smallest safely representable integer value. */
export const MIN_SAFE_INTEGER: f64;
/** Largest safely representable integer value. */
export const MAX_SAFE_INTEGER: f64;
/** Difference between 1 and the smallest representable value greater than 1. */
export const EPSILON: f64;
/** Loads a 64-bit float from memory. */
export function load(offset: usize, constantOffset?: usize): f64;
/** Stores a 64-bit float to memory. */
export function store(offset: usize, value: f64, constantOffset?: usize): void;
}
// Built-ins
/** Performs the sign-agnostic count leading zero bits operation on a 32-bit or 64-bit integer. All zero bits are considered leading if the value is zero. */
declare function clz<T = i32 | i64>(value: T): T;
/** Performs the sign-agnostic count tailing zero bits operation on a 32-bit or 64-bit integer. All zero bits are considered trailing if the value is zero. */
declare function ctz<T = i32 | i64>(value: T): T;
/** Performs the sign-agnostic count number of one bits operation on a 32-bit or 64-bit integer. */
declare function popcnt<T = i32 | i64>(value: T): T;
/** Performs the sign-agnostic rotate left operation on a 32-bit or 64-bit integer. */
declare function rotl<T = i32 | i64>(value: T, shift: T): T;
/** Performs the sign-agnostic rotate right operation on a 32-bit or 64-bit integer. */
declare function rotr<T = i32 | i64>(value: T, shift: T): T;
/** Computes the absolute value of an integer or float. */
declare function abs<T = i32 | i64 | f32 | f64>(value: T): T;
/** Determines the maximum of two integers or floats. If either operand is `NaN`, returns `NaN`. */
declare function max<T = i32 | i64 | f32 | f64>(left: T, right: T): T;
/** Determines the minimum of two integers or floats. If either operand is `NaN`, returns `NaN`. */
declare function min<T = i32 | i64 | f32 | f64>(left: T, right: T): T;
/** Performs the ceiling operation on a 32-bit or 64-bit float. */
declare function ceil<T = f32 | f64>(value: T): T;
/** Composes a 32-bit or 64-bit float from the magnitude of `x` and the sign of `y`. */
declare function copysign<T = f32 | f64>(x: T, y: T): T;
/** Performs the floor operation on a 32-bit or 64-bit float. */
declare function floor<T = f32 | f64>(value: T): T;
/** Rounds to the nearest integer tied to even of a 32-bit or 64-bit float. */
declare function nearest<T = f32 | f64>(value: T): T;
/** Reinterprets the bits of the specified value as type `T`. Valid reinterpretations are u32/i32 to/from f32 and u64/i64 to/from f64. */
declare function reinterpret<T = i32 | i64 | f32 | f64>(value: number): T;
/** Selects one of two pre-evaluated values depending on the condition. */
declare function select<T>(ifTrue: T, ifFalse: T, condition: bool): T;
/** Calculates the square root of a 32-bit or 64-bit float. */
declare function sqrt<T = f32 | f64>(value: T): T;
/** Rounds to the nearest integer towards zero of a 32-bit or 64-bit float. */
declare function trunc<T = f32 | f64>(value: T): T;
/** Loads a value of the specified type from memory. Equivalent to dereferncing a pointer in other languages. */
declare function load<T>(ptr: usize, constantOffset?: usize): T;
/** Stores a value of the specified type to memory. Equivalent to dereferencing a pointer in other languages when assigning a value. */
declare function store<T>(ptr: usize, value: any, constantOffset?: usize): void;
/** Returns the current memory size in units of pages. One page is 64kb. */
declare function current_memory(): i32;
/** Grows linear memory by a given unsigned delta of pages. One page is 64kb. Returns the previous memory size in units of pages or `-1` on failure. */
declare function grow_memory(value: i32): i32;
/** Copies n bytes from the specified source to the specified destination in memory. These regions may overlap. */
declare function move_memory(destination: usize, source: usize, n: usize): void;
/** Sets n bytes beginning at the specified destination in memory to the specified byte value. */
declare function set_memory(destination: usize, value: u8, count: usize): void;
/** Compares two chunks of memory. Returns `0` if equal, otherwise the difference of the first differing bytes. */
declare function compare_memory(vl: usize, vr: usize, n: usize): i32;
/** Allocates a chunk of memory of the specified size and returns a pointer to it. */
declare function allocate_memory(size: usize): usize;
/** Disposes a chunk of memory by its pointer. */
declare function free_memory(ptr: usize): void;
/** Emits an unreachable operation that results in a runtime error when executed. Both a statement and an expression of any type. */
declare function unreachable(): any; // sic
/** [Polyfill] Performs the sign-agnostic reverse bytes **/
declare function bswap<T = i8 | u8 | i16 | u16 | i32 | u32 | i64 | u64 | isize | usize>(value: T): T;
/** [Polyfill] Performs the sign-agnostic reverse bytes only for last 16-bit **/
declare function bswap16<T = i8 | u8 | i16 | u16 | i32 | u32>(value: T): T;
/** NaN (not a number) as a 32-bit or 64-bit float depending on context. */
declare const NaN: f32 | f64;
/** Positive infinity as a 32-bit or 64-bit float depending on context. */
declare const Infinity: f32 | f64;
/** Heap base offset. */
declare const HEAP_BASE: usize;
/** Determines the byte size of the specified underlying core type. Compiles to a constant. */
declare function sizeof<T>(): usize;
/** Determines the alignment (log2) of the specified underlying core type. Compiles to a constant. */
declare function alignof<T>(): usize;
/** Determines the offset of the specified field within the given class type. Returns the class type's end offset if field name has been omitted. Compiles to a constant. */
declare function offsetof<T>(fieldName?: string): usize;
/** Changes the type of any value of `usize` kind to another one of `usize` kind. Useful for casting class instances to their pointer values and vice-versa. Beware that this is unsafe.*/
declare function changetype<T>(value: any): T;
/** Explicitly requests no bounds checks on the provided expression. Useful for array accesses. */
declare function unchecked<T>(value: T): T;
/** Emits a `call_indirect` instruction, calling the specified function in the function table by index with the specified arguments. Does result in a runtime error if the arguments do not match the called function. */
declare function call_indirect<T>(target: Function | u32, ...args: any[]): T;
/** Tests if a 32-bit or 64-bit float is `NaN`. */
declare function isNaN<T = f32 | f64>(value: T): bool;
/** Tests if a 32-bit or 64-bit float is finite, that is not `NaN` or +/-`Infinity`. */
declare function isFinite<T = f32 | f64>(value: T): bool;
/** Tests if the specified type *or* expression is of an integer type and not a reference. Compiles to a constant. */
declare function isInteger<T>(value?: any): value is number;
/** Tests if the specified type *or* expression is of a float type. Compiles to a constant. */
declare function isFloat<T>(value?: any): value is number;
/** Tests if the specified type *or* expression can represent negative numbers. Compiles to a constant. */
declare function isSigned<T>(value?: any): value is number;
/** Tests if the specified type *or* expression is of a reference type. Compiles to a constant. */
declare function isReference<T>(value?: any): value is object | string;
/** Tests if the specified type *or* expression can be used as a string. Compiles to a constant. */
declare function isString<T>(value?: any): value is string | String;
/** Tests if the specified type *or* expression can be used as an array. Compiles to a constant. */
declare function isArray<T>(value?: any): value is Array<any>;
/** Traps if the specified value is not true-ish, otherwise returns the (non-nullable) value. */
declare function assert<T>(isTrueish: T, message?: string): T & object; // any better way to model `: T != null`?
/** Parses an integer string to a 64-bit float. */
declare function parseInt(str: string, radix?: i32): f64;
/** Parses an integer string to a 32-bit integer. */
declare function parseI32(str: string, radix?: i32): i32;
/** Parses an integer string to a 64-bit integer. */
declare function parseI64(str: string, radix?: i32): i64;
/** Parses a string to a 64-bit float. */
declare function parseFloat(str: string): f64;
/** Returns the 64-bit floating-point remainder of `x/y`. */
declare function fmod(x: f64, y: f64): f64;
/** Returns the 32-bit floating-point remainder of `x/y`. */
declare function fmodf(x: f32, y: f32): f32;
// Standard library
/** Class representing a generic, fixed-length raw binary data buffer. */
declare class ArrayBuffer {
/** The size, in bytes, of the array. */
readonly byteLength: i32;
/** Constructs a new array buffer of the given length in bytes. */
constructor(length: i32);
/** Returns a copy of this array buffer's bytes from begin, inclusive, up to end, exclusive. */
slice(begin?: i32, end?: i32): ArrayBuffer;
}
/** Interface for a typed view on an array buffer. */
interface ArrayBufferView<T> {
[key: number]: T;
/** The {@link ArrayBuffer} referenced by this view. */
readonly buffer: ArrayBuffer;
/** The offset in bytes from the start of the referenced {@link ArrayBuffer}. */
readonly byteOffset: i32;
/** The length in bytes from the start of the referenced {@link ArrayBuffer}. */
readonly byteLength: i32;
}
/* @internal */
declare abstract class TypedArray<T> implements ArrayBufferView<T> {
[key: number]: T;
/** Number of bytes per element. */
static readonly BYTES_PER_ELEMENT: usize;
/** Constructs a new typed array. */
constructor(length: i32);
/** The {@link ArrayBuffer} referenced by this view. */
readonly buffer: ArrayBuffer;
/** The offset in bytes from the start of the referenced {@link ArrayBuffer}. */
readonly byteOffset: i32;
/** The length in bytes from the start of the referenced {@link ArrayBuffer}. */
readonly byteLength: i32;
/** The length (in elements). */
readonly length: i32;
/** Returns a new TypedArray of this type on the same ArrayBuffer from begin inclusive to end exclusive. */
subarray(begin?: i32, end?: i32): this;
}
/** An array of twos-complement 8-bit signed integers. */
declare class Int8Array extends TypedArray<i8> {}
/** An array of 8-bit unsigned integers. */
declare class Uint8Array extends TypedArray<u8> {}
/** An array of twos-complement 16-bit signed integers. */
declare class Int16Array extends TypedArray<i16> {}
/** An array of 16-bit unsigned integers. */
declare class Uint16Array extends TypedArray<u16> {}
/** An array of twos-complement 32-bit signed integers. */
declare class Int32Array extends TypedArray<i32> {}
/** An array of 32-bit unsigned integers. */
declare class Uint32Array extends TypedArray<u32> {}
/** An array of twos-complement 64-bit signed integers. */
declare class Int64Array extends TypedArray<i64> {}
/** An array of 64-bit unsigned integers. */
declare class Uint64Array extends TypedArray<u64> {}
/** An array of 32-bit floating point numbers. */
declare class Float32Array extends TypedArray<f32> {}
/** An array of 64-bit floating point numbers. */
declare class Float64Array extends TypedArray<f64> {}
/** Class representing a sequence of values of type `T`. */
declare class Array<T> {
[key: number]: T;
/** Current length of the array. */
length: i32;
/** Constructs a new array. */
constructor(capacity?: i32);
every(callbackfn: (element: T, index: i32, array?: Array<T>) => bool): bool;
findIndex(predicate: (element: T, index: i32, array?: Array<T>) => bool): i32;
includes(searchElement: T, fromIndex?: i32): bool;
indexOf(searchElement: T, fromIndex?: i32): i32;
lastIndexOf(searchElement: T, fromIndex?: i32): i32;
push(element: T): void;
pop(): T;
forEach(callbackfn: (value: T, index: i32, array: Array<T>) => void): void;
map<U>(callbackfn: (value: T, index: i32, array: Array<T>) => U): Array<U>;
filter(callbackfn: (value: T, index: i32, array: Array<T>) => bool): Array<T>;
reduce<U>(callbackfn: (previousValue: U, currentValue: T, currentIndex: i32, array: Array<T>) => U, initialValue: U): U;
reduceRight<U>(callbackfn: (previousValue: U, currentValue: T, currentIndex: i32, array: Array<T>) => U, initialValue: U): U;
shift(): T;
some(callbackfn: (element: T, index: i32, array?: Array<T>) => bool): bool;
unshift(element: T): i32;
slice(from: i32, to?: i32): T[];
splice(start: i32, deleteCount?: i32): void;
reverse(): T[];
sort(comparator?: (a: T, b: T) => i32): this;
}
/** Class representing a C-like array of values of type `T` with limited capabilities. */
declare class CArray<T> {
[key: number]: T;
private constructor();
}
/** Class representing a sequence of characters. */
declare class String {
static fromCharCode(ls: i32, hs?: i32): string;
static fromCharCodes(arr: u16[]): string;
static fromCodePoint(cp: i32): string;
static fromCodePoints(arr: i32[]): string;
readonly length: i32;
readonly lengthUTF8: i32;
charAt(index: u32): string;
charCodeAt(index: u32): u16;
concat(other: string): string;
endsWith(other: string): bool;
indexOf(other: string): u32;
includes(other: string): bool;
startsWith(other: string): bool;
substr(start: u32, length?: u32): string;
substring(start: u32, end?: u32): string;
trim(): string;
trimLeft(): string;
trimRight(): string;
repeat(count?: i32): string;
toString(): string;
toUTF8(): usize;
}
/** Class for representing a runtime error. Base class of all errors. */
declare class Error {
/** Error name. */
name: string;
/** Message provided on construction. */
message: string;
/** Stack trace. */
stack: string;
/** Constructs a new error, optionally with a message. */
constructor(message?: string);
}
/** Class for indicating an error when a value is not in the set or range of allowed values. */
declare class RangeError extends Error { }
interface Boolean {}
interface Function {}
interface IArguments {}
interface Number {}
interface Object {}
interface RegExp {}
declare class Set<T> {
readonly size: i32;
has(value: T): bool;
add(value: T): void;
delete(value: T): bool;
clear(): void;
}
interface IMath<T> {
/** The base of natural logarithms, e, approximately 2.718. */
readonly E: T;
/** The natural logarithm of 2, approximately 0.693. */
readonly LN2: T;
/** The natural logarithm of 10, approximately 2.302. */
readonly LN10: T;
/** The base 2 logarithm of e, approximately 1.442. */
readonly LOG2E: T;
/** The base 10 logarithm of e, approximately 0.434. */
readonly LOG10E: T;
/** The ratio of the circumference of a circle to its diameter, approximately 3.14159. */
readonly PI: T;
/** The square root of 1/2, approximately 0.707. */
readonly SQRT1_2: T;
/** The square root of 2, approximately 1.414. */
readonly SQRT2: T;
/** Returns the absolute value of `x`. */
abs(x: T): T;
/** Returns the arccosine (in radians) of `x`. */
acos(x: T): T;
/** Returns the hyperbolic arc-cosine of `x`. */
acosh(x: T): T;
/** Returns the arcsine (in radians) of `x` */
asin(x: T): T;
/** Returns the hyperbolic arcsine of `x`. */
asinh(x: T): T;
/** Returns the arctangent (in radians) of `x`. */
atan(x: T): T;
/** Returns the arctangent of the quotient of its arguments. */
atan2(y: T, x: T): T;
/** Returns the hyperbolic arctangent of `x`. */
atanh(x: T): T;
/** Returns the cube root of `x`. */
cbrt(x: T): T;
/** Returns the smallest integer greater than or equal to `x`. */
ceil(x: T): T;
/** Returns the number of leading zero bits in the 32-bit binary representation of `x`. */
clz32(x: T): T;
/** Returns the cosine (in radians) of `x`. */
cos(x: T): T;
/** Returns the hyperbolic cosine of `x`. */
cosh(x: T): T;
/** Returns e to the power of `x`. */
exp(x: T): T;
/** Returns e to the power of `x`, minus 1. */
expm1(x: T): T;
/** Returns the largest integer less than or equal to `x`. */
floor(x: T): T;
/** Returns the nearest 32-bit single precision float representation of `x`. */
fround(x: T): f32;
/** Returns the square root of the sum of squares of its arguments. */
hypot(value1: T, value2: T): T; // TODO: rest
/** Returns the result of the C-like 32-bit multiplication of `a` and `b`. */
imul(a: T, b: T): T;
/** Returns the natural logarithm (base e) of `x`. */
log(x: T): T;
/** Returns the base 10 logarithm of `x`. */
log10(x: T): T;
/** Returns the natural logarithm (base e) of 1 + `x`. */
log1p(x: T): T;
/** Returns the base 2 logarithm of `x`. */
log2(x: T): T;
/** Returns the largest-valued number of its arguments. */
max(value1: T, value2: T): T; // TODO: rest
/** Returns the lowest-valued number of its arguments. */
min(value1: T, value2: T): T; // TODO: rest
/** Returns `base` to the power of `exponent`. */
pow(base: T, exponent: T): T;
/** Returns a pseudo-random number in the range from 0.0 inclusive up to but not including 1.0. */
random(): T;
/** Returns the value of `x` rounded to the nearest integer. */
round(x: T): T;
/** Returns the sign of `x`, indicating whether the number is positive, negative or zero. */
sign(x: T): T;
/** Returns the sine of `x`. */
sin(x: T): T;
/** Returns the hyperbolic sine of `x`. */
sinh(x: T): T;
/** Returns the square root of `x`. */
sqrt(x: T): T;
/** Returns the tangent of `x`. */
tan(x: T): T;
/** Returns the hyperbolic tangent of `x`. */
tanh(x: T): T;
/** Returns the integer part of `x` by removing any fractional digits. */
trunc(x: T): T;
}
interface INativeMath<T> extends IMath<T> {
/** Seeds the random number generator. */
seedRandom(value: i64): void;
/** Returns the floating-point remainder of `x / y` (rounded towards zero). */
mod(x: T, y: T): T;
/** Returns the floating-point remainder of `x / y` (rounded to nearest). */
rem(x: T, y: T): T;
}
/** Double precision math imported from JavaScript. */
declare const JSMath: IMath<f64>;
/** Double precision math implemented natively. */
declare const NativeMath: INativeMath<f64>;
/** Single precision math implemented natively. */
declare const NativeMathf: INativeMath<f32>;
/** Alias of {@link NativeMath} or {@link JSMath} respectively. Defaults to `NativeMath`. */
declare const Math: IMath<f64>;
/** Alias of {@link NativeMathf} or {@link JSMath} respectively. Defaults to `NativeMathf`. */
declare const Mathf: IMath<f32>;
// Internal decorators
/** Annotates an element as a program global. */
declare function global(target: Function, propertyKey: string, descriptor: any): void;
/** Annotates a method as an operator overload for the specified `token`. */
declare function operator(token: string): (target: any, propertyKey: string, descriptor: any) => void;
/** Annotates a class as being unmanaged with limited capabilities. */
declare function unmanaged(target: Function): any;
/** Annotates a class as being sealed / non-derivable. */
declare function sealed(target: Function): any;
/** Annotates a method or function as always inlined. */
declare function inline(target: any, propertyKey: any, descriptor: any): any;