💫 Error Handling

In nyx, every executor holds an ErrorHandler, where errors thrown by middlewares and executed objects are handled.

An error handler behaves like a map of consumers, where the key is the error constructor (the class) and the value is an ErrorConsumer, which is a callback that receives the error, the object, the arguments that were used to call it and the bot.

🔨 Manipulating consumers

To add a consumer, use ErrorHandler#setConsumer():

const errorHandler = bot.getCommandManager().getExecutor().getErrorHandler();

errorHandler.setConsumer(
  MyError,
  (
    error: MyError,
    command: ExecutableCommand<CommandData>,
    args: CommandExecutionArgs,
  ) => {
    const [meta] = args;
    const bot = meta.getBot(true);

    const { name } = command.getData();

    logger.error(
      `MyError occurred while executing ${name} with args: ${args}`,
      error
    );
  }
);

To remove a consumer, use ErrorHandler#removeConsumer():

const errorHandler = bot.getCommandManager().getExecutor().getErrorHandler();

errorHandler.removeConsumer(MyError);

⏬ Fallback consumer

The ErrorHandler has a fallback consumer, which is used if no consumer was found for a given error.

const errorHandler = bot.getCommandManager().getExecutor().getErrorHandler();

errorHandler.setFallbackConsumer(
  (
    error: object,
    command: ExecutableCommand<CommandData>,
    args: CommandExecutionArgs,
  ) => {
    const [meta] = args;
    const bot = meta.getBot(true);
    const { name } = command.getData();

    logger.error(
      `An unhandled error occurred while executing ${name} with args: ${args}`,
      error
    );
  }
);

🔎 Consumer resolution

The BasicErrorHandler resolves a consumer based on the error prototype, going from the top to the bottom.

note

For instance, if MyError2 extends MyError1 which extends MyError, then it will first try to resolve the consumer for MyError2, then for MyError1 and finally for MyError.

If no consumer is found, the fallback consumer is used, whose error type is simply object.