Programming to Exceptions

Read the articles

Read Parts 1, 2, and 3 in order: method contracts, exception design, then logging and progress. The concepts apply across languages; the worked examples use C#.

• 

  • Error Handling
  • Method Design
  • Boundary Validation

  Programming to Exceptions, Part 1: Method Contracts and Failure

  Action methods do their work or throw; query methods return what they promise or don't return. Failures propagate to one catch at the entry point, and each door validates incoming data.

  28 October 2019Markdown

• 

  • Error Handling
  • Structured Logging
  • Class Design

  Programming to Exceptions, Part 2: Diagnostics and Boundaries

  One log record should explain a failure to the support team. Exceptions carry a message, reason, action and context, are translated where meaning is known, and are logged once.

  28 October 2019Markdown

• 

  • Error Handling
  • Structured Logging
  • Architecture

  Programming to Exceptions, Part 3: Logging and Progress

  Each failure is logged once at the entry point, with its facts as Application Insights custom dimensions for KQL queries and alerts. A shared progress logger reports long-running work.

  28 October 2019Markdown