Skip to content

TESTING.md

Latest commit

 

History

History
50 lines (42 loc) · 2.83 KB

TESTING.md

File metadata and controls

50 lines (42 loc) · 2.83 KB
Raw

Testing

Run ./gradlew test (or sh ./gradlew test) with JDK 25. ./gradlew check includes the same suite. Reports are in build/reports/tests/test/index.html. CI runs check in a separate Test job; container builds/publication depend on it.

Scope

JUnit Jupiter and Mockito test our decisions, not the external services:

  • GuildAudioTest: immediate playback versus FIFO queueing, skip/stop/pause, exhausted or failed next-track handling, track-end policies, and advancement despite failed error notifications.
  • AudioManagerTest: one search fallback, direct playlists versus search results, request metadata, queue outcomes, failure cleanup without interrupting existing playback, and voice connection policy.
  • YoutubeConfigurationTest: pure environment parsing, blank/trimmed values, independent optional settings, paired poToken/visitor data, and credential redaction in parsed settings' string representation and validation errors.
  • TimerTest: default and interval notice schedules, one-minute warning deduplication, mentions, guild-only invocation, and completion-time voice return for connected members only, Administrator defaults without caller enforcement, and bot-permission preflight (text, destination voice, source voice/stage, and non-mentionable roles). Scheduled notices and completion callbacks are captured without waiting or contacting Discord.

Tests inject the loader and activity callback, manually deliver load/playback callbacks, and parse configuration from a map. They never construct a connected bot or authenticate with YouTube. Mockito is supplied as an explicit test JVM agent for modern JDKs; no application runtime dependency is added.

Keeping this lightweight

  • Add tests for plausible bugs in application behavior, not a coverage target.
  • Prefer parameterized boundary cases and observable outcomes. Interaction assertions are appropriate for decisions such as disconnecting or advancing exactly once, not incidental builder call order.
  • No sleeps, network, credentials, real audio, or whole-bot startup in tests.
  • Do not test Discord delivery/voice transport, YouTube availability, cipher or OAuth refresh, decoding, library internals, reflection, generated accessors, trivial command delegation, or exact prose/embed styling.
  • Keep fixtures local and small; do not build a Discord/Lavaplayer simulator.
  • Add a focused regression test when fixing a substantive bug.

After deployment or dependency updates, manually check Discord connectivity, playback, queue advancement, pause, and stop on the intended host. Also create a short /timer with updates, a mention, and a voice-return destination; verify notice delivery and voice movement with the bot's actual channel permissions. YouTube/IP, cipher, and voice transport issues belong to this operational smoke check, not the deterministic Gradle suite.