Timedora

Classifying Application Failures

A practical introduction to distinguishing application crashes from hangs and using timing and user activity to narrow possible causes.

, , and Context

Tell a from a

A means the application terminates unexpectedly. A means it is still running but no longer responds. Record what you actually observe: does the app exit, freeze, restart, or does the operating system terminate it? These behaviors are useful clues, so avoid describing every failure simply as a “.”

Use timing and activity as clues

A during startup often points toward initialization, configuration, permissions, or dependency problems. A during normal use may be tied to a particular action or input, workload, or resource condition. These are clues for narrowing the investigation, not proof of a specific cause.

For each incident, note when it happened, any error message, the application version and operating system, and what the user was doing. As a hypothetical application of this advice, recording that an app froze while opening a particular file is more informative than noting only that “the app failed.” The observed behavior and its context help guide what to investigate next.