fix(runtime): respect keepPlaying option in player seek

[calcarazgre646]

Summary

Closes the follow-up I promised in #842 (comment): the __player runtime adapter ignored the { keepPlaying: true } option that A/E Jump-to-in/out shortcuts send, so playback pauses on every press even though the caller asked to preserve it. Only the wrapTimeline (GSAP) path honoured the option.

Background

PR #842 added seek: (time, options?: { keepPlaying?: boolean }) to PlaybackAdapter and wired the A/E shortcuts in usePlaybackKeyboard to pass { keepPlaying: true }. The fix landed correctly for compositions backed by __timeline / __timelines (GSAP), because wrapTimeline reads the option and skips its internal tl.pause().

For compositions backed by window.__player (packages/core/src/runtime/init.ts:1529), useTimelinePlayer.getAdapter() returns the runtime adapter directly without a wrapper. That adapter's seek is basePlayer.seek from createRuntimePlayer (packages/core/src/runtime/player.ts:147), and the implementation was:

seek: (timeSeconds: number) => {
  // ...quantize + deterministic seek...
  deps.setIsPlaying(false);
  deps.onSyncMedia(quantized, false);
  // ...
},

setIsPlaying(false) runs unconditionally, so A/E pauses the moment the shortcut fires. The option that the Studio sends never reaches the implementation because both the local basePlayer shape in createPlayerApiCompat and the public PlayerAPI.seek / RuntimePlayer.seek types only declared (time: number) => void.

Fix

1. Extend the type contract from end to end so the option is preserved through createPlayerApiCompat:

   • core.types.ts — PlayerAPI.seek(time, options?: { keepPlaying?: boolean })
   • runtime/types.ts — RuntimePlayer.seek signature
   • runtime/init.ts:88 — local basePlayer shape in createPlayerApiCompat

2. Implement the behaviour in createRuntimePlayer.seek:

   • Capture wasPlaying before the deterministic seek helper pauses the master and rearmed siblings.
   • When options.keepPlaying && wasPlaying, resume the master + siblings, reapply playbackRate via timeScale, and emit onDeterministicPlay + onShowNativeVideos + onSyncMedia(t, true) so media and analytics stay in sync.
   • Default branch (no options, explicit keepPlaying: false, or wasPlaying === false) keeps the existing setIsPlaying(false) + onSyncMedia(t, false) path. No back-compat change.

Test plan

Added 6 new tests in packages/core/src/runtime/player.test.ts under describe(\"keepPlaying option\"):

• Preserves play state when keepPlaying: true and playback was active (no setIsPlaying(false), emits onDeterministicPlay + onSyncMedia(t, true)).
• Resumes the master timeline after the deterministic seek pauses it (asserts play() is invoked after the last internal pause() via invocationCallOrder).
• Applies playbackRate to master and siblings on resume.
• Stays paused when keepPlaying: true but playback was not active (intent is preserve, not force).
• Explicit keepPlaying: false matches default behaviour.
• No options matches default behaviour (back-compat).

Other validations:

• bun run --cwd packages/core test → 868/868 passing (was 862, +6 new).
• bun run --cwd packages/studio test → 514/514 passing, no regression in the consuming side.
• bun run --cwd packages/core typecheck and bun run --cwd packages/studio typecheck clean.
• bunx oxlint and bunx oxfmt --check clean on touched files.
• Lefthook pre-commit ran lint + format + typecheck + commitlint.

Out of scope

createStaticSeekPlaybackAdapter in packages/studio/src/player/lib/playbackAdapter.ts still has seek: (time) => void (no options). Its current behaviour happens to preserve playback by accident (the playing flag drives the RAF ticker independently), but the contract is mismatched. Happy to follow up if a maintainer wants the static-seek adapter brought in line with the same shape.

Related

• PR fix(studio): preserve playback state on Jump-to-in/out shortcuts #842 (the original fix, GSAP path).
• jrusso's review on fix(studio): preserve playback state on Jump-to-in/out shortcuts #842 raising the __player path concern.
• Issue (studio) bugs in I/O markers feature #834 (the originating I/O markers bug list).