dispatch

Dispatch to a native add-on applying a unary function to an input strided array.

Usage

var dispatch = require( '@stdlib/strided/base/unary-addon-dispatch' );

dispatch( addon, fallback )

Returns a function which dispatches to a native add-on applying a unary function to an input strided array.

function addon( N, dtypeX, x, strideX, dtypeY, y, strideY ) {
    // Call into native add-on...
}

function fallback( N, dtypeX, x, strideX, dtypeY, y, strideY ) {
    // Fallback JavaScript implementation...
}

// Create a dispatch function:
var f = dispatch( addon, fallback );

// ...

// Invoke the dispatch function with strided array arguments:
f( 2, 'generic', [ 1, 2 ], 1, 'generic', [ 0, 0 ], 1 );

The returned function has the following signature:

f( N, dtypeX, x, strideX, dtypeY, y, strideY )

where

  • N: number of indexed elements.
  • dtypeX: x data type.
  • x: input array.
  • strideX: x stride length.
  • dtypeY: y data type.
  • y: output array.
  • strideY: y stride length.

The addon function should have the following signature:

f( N, dtypeX, x, strideX, dtypeY, y, strideY )

where

  • N: number of indexed elements.
  • dtypeX: x data type (enumeration constant).
  • x: input array.
  • strideX: x stride length.
  • dtypeY: y data type (enumeration constant).
  • y: output array.
  • strideY: y stride length.

The fallback function should have the following signature:

f( N, dtypeX, x, strideX, dtypeY, y, strideY )

where

  • N: number of indexed elements.
  • dtypeX: x data type.
  • x: input array.
  • strideX: x stride length.
  • dtypeY: y data type.
  • y: output array.
  • strideY: y stride length.

dispatch.ndarray( addon, fallback )

Returns a function which dispatches to a native add-on applying a unary function to an input strided array using alternative indexing semantics.

function addon( N, dtypeX, x, strideX, dtypeY, y, strideY ) {
    // Call into native add-on...
}

function fallback( N, dtypeX, x, strideX, offsetX, dtypeY, y, strideY, offsetY ) {
    // Fallback JavaScript implementation...
}

// Create a dispatch function:
var f = dispatch.ndarray( addon, fallback );

// ...

// Invoke the dispatch function with strided array arguments:
f( 2, 'generic', [ 1, 2 ], 1, 0, 'generic', [ 0, 0 ], 1, 0 );

The returned function has the following signature:

f( N, dtypeX, x, strideX, offsetX, dtypeY, y, strideY, offsetY )

where

  • N: number of indexed elements.
  • dtypeX: x data type.
  • x: input array.
  • strideX: x stride length.
  • offsetX: starting x index.
  • dtypeY: y data type.
  • y: output array.
  • strideY: y stride length.
  • offsetY: starting y index.

The addon function should have the following signature:

f( N, dtypeX, x, strideX, dtypeY, y, strideY )

where

  • N: number of indexed elements.
  • dtypeX: x data type (enumeration constant).
  • x: input array.
  • strideX: x stride length.
  • dtypeY: y data type (enumeration constant).
  • y: output array.
  • strideY: y stride length.

The fallback function should have the following signature:

f( N, dtypeX, x, strideX, offsetX, dtypeY, y, strideY, offsetY )

where

  • N: number of indexed elements.
  • dtypeX: x data type.
  • x: input array.
  • strideX: x stride length.
  • offsetX: starting x index.
  • dtypeY: y data type.
  • y: output array.
  • strideY: y stride length.
  • offsetY: starting y index.

Notes

  • To determine whether to dispatch to the addon function, the returned dispatch function checks whether the provided arrays are typed arrays. If the provided arrays are typed arrays, the dispatch function invokes the addon function; otherwise, the dispatch function invokes the fallback function.

Examples

var Float64Array = require( '@stdlib/array/float64' );
var dispatch = require( '@stdlib/strided/base/unary-addon-dispatch' );

function addon( N, dtypeX, x, strideX, dtypeY, y, strideY ) {
    console.log( x );
    // => <Float64Array>[ 3, 4 ]

    console.log( y );
    // => <Float64Array>[ 7, 8 ]
}

function fallback( N, dtypeX, x, strideX, offsetX, dtypeY, y, strideY, offsetY ) {
    console.log( x );
    // => [ 1, 2, 3, 4 ]

    console.log( y );
    // => [ 5, 6, 7, 8 ]
}

// Create a dispatch function:
var f = dispatch.ndarray( addon, fallback );

// Create strided arrays:
var x = new Float64Array( [ 1, 2, 3, 4 ] );
var y = new Float64Array( [ 5, 6, 7, 8 ] );

// Dispatch to the add-on function:
f( 2, 'float64', x, 1, 2, 'float64', y, 1, 2 );

// Define new strided arrays:
x = [ 1, 2, 3, 4 ];
y = [ 5, 6, 7, 8 ];

// Dispatch to the fallback function:
f( 2, 'generic', x, 1, 2, 'generic', y, 1, 2 );
Did you find this page helpful?