Skip to content

Latest commit

 

History

History

input-selection

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Cardano JS SDK | Input Selection | CIP-0002

This package implements concepts from the draft specification being developed in CIP-0002.

Currently, there is only 1 input selection algorithm: RoundRobinRandomImprove, which is an Random-Improve adaptation that handles asset selection.

Usage Example

import { roundRobinRandomImprove, defaultSelectionConstraints, InputSelector, SelectionResult, SelectionSkeleton } from '@cardano-sdk/input-selection';
import { loadCardanoSerializationLib, CSL, CardanoSerializationLib, ProtocolParametersRequiredByWallet } from '@cardano-sdk/core';

const demo = async (protocolParameters: ProtocolParametersRequiredByWallet): Promise<SelectionResult> => {
  const csl: CardanoSerializationLib = await loadCardanoSerializationLib();
  const selector: InputSelector = roundRobinRandomImprove(coinsPerUtxoWord);
  // It is important that you use the same instance of cardano-serialization-lib across your application.
  // Bad: TransactionUnspentOutput.new(...)
  // Good: CSL.TransactionUnspentOutput.new(...)
  const utxo: CSL.TransactionUnspentOutput[] = [CSL.TransactionUnspentOutput.new(...), ...];
  const outputs: CSL.TransactionOutput[] = [CSL.TransactionOutput.new(...), ...];
  // Used to estimate min fee and validate transaction size
  const buildTx = (inputSelection: SelectionSkeleton): Promise<CSL.Transaction> => {...};
  const constraints = defaultSelectionConstraints({
    protocolParameters, buildTx,
  });

  return selector.select({
    utxo,
    outputs,
    constraints,
  });
};

Tests

Input selection is tested with property-based tests using fast-check, as well as a few regular example-based tests.

Due to nature of property-based tests, code coverage report is slightly different on each build.

RoundRobinRandomImprove has 100% code coverage when using high numRuns option (e.g. 100_000).

Note that to run it with high numRuns you need to increase Jest and fast-check timeout.

See code coverage report.