This module provides behaviour for cancel phase error handling.

When a critical error occurs during the Cancel phase (exception raised, exit signal, or throw), the default behavior is to raise the error. However, you can register a custom error handler to implement alternative recovery strategies.

Recovery Strategies

Some solutions you might want to implement:

• Send notification to a Slack channel about need of manual resolution
• Retry the cancel operation with different parameters
• Log the error and continue with other cancellations
• Spawn a new supervised process that would retry cancellation later (useful when you have connection issues that would be resolved at some point)

Example

defmodule MyCancelErrorHandler do
  @behaviour TCC.CancelErrorHandler

  require Logger

  @impl true
  def handle_error({:exception, exception, stacktrace}, cancellations_to_run, opts) do
    Logger.error("Cancel failed: #{Exception.message(exception)}")
    # Notify external system
    notify_slack(exception, cancellations_to_run)
    {:error, :cancel_failed}
  end

  def handle_error({:exit, reason}, cancellations_to_run, opts) do
    Logger.error("Cancel exited: #{inspect(reason)}")
    {:error, :cancel_exit}
  end

  def handle_error({:throw, error}, cancellations_to_run, opts) do
    Logger.error("Cancel threw: #{inspect(error)}")
    {:error, :cancel_throw}
  end
end

TCC.new()
|> TCC.with_cancel_error_handler(MyCancelErrorHandler)
|> TCC.run(:payment, &try_fn/2, &confirm_fn/2, &cancel_fn/2)
|> TCC.execute(%{})

Important Notes

• The error handler is only called for critical errors (exceptions, exits, throws)
• Regular {:error, reason} returns from cancel functions are handled normally
• The handler receives information about remaining cancellations that need to run
• You should always return {:error, reason} from the handler