Testing

Bot tasks should be fully tested.

A required step in bot development is testing. Fully test all bot tasks before they are deployed to production. The goal is to identify and correct known errors, and prevent unexpected events from causing the bot to fail. If a bot does not pass testing:

  • Correct the error

  • Put error handling processes in place

Error handling web bots

Creating bots for browser-based applications can be a moving target. If a web page changes it often breaks the bot. If a web page does not appear when the bot expects, it can cause an error. The key to a successful bot is predicting and handling unexpected events.

When creating bots that rely on network connectivity, ensure to include steps for gracefully handling networking connectivity issues. If your bot works with a web browser it needs to gracefully handle when the browser cannot load the target page due to an outage.

For example, consider a bot that downloads a file from a web site. After clicking on the download link, the bot waits 15 seconds for the download prompt to appear in the browser. The bot uses a wait for window to exist command to determine when the Save As dialog appears.

To debug your bots, see Debugging a task.