↓ Ir para o conteúdo principal

← todas as notas

📎 Webclip

There Are No Bugs, Just TODOs

The essay argues that issue trackers work best when they stay small and operational. A ticket should have a clear owner, belong to one ordered queue, move through a simple state flow, and be broken down when it spans more than one deployment. It also calls for a separate channel for internal support requests so the main tracker does not turn into a swamp.

Reading notes
#

  • Issue trackers are useful, but they should not replace understanding the underlying software delivery.
  • A ticket needs to reach someone who can do the work, with triage handled reliably when ownership is unclear.
  • Every team should have one ordered queue, and all incoming work should be prioritized against each other in one place.
  • The simplest useful ticket states are Todo, Doing, and Done, with extra states added only when needed.
  • A ticket should map to a single deployment, and larger work should be split into sub-tickets with the relationship recorded.
  • Absolute priority fields create abuse; priority should be relative to other issues.
  • Ticket type often creates unproductive debate, because the team should focus on resolving the expectation mismatch.
  • Software version tracking makes sense for shipped libraries and on-premise packages, but not for continuously delivered SaaS.
  • Severity often duplicates priority and depends on role and effort.
  • The system should make it easy to close issues aggressively, or it will turn into an unmanageable ticket swamp.
  • Questions for the development team need a separate system, because random messages and badly specified tickets overload the queue.