Skip to content

fix(test): set burst start at composition begin for IME input (@boergeson)#7760

Open
boergeson wants to merge 1 commit intomonkeytypegame:masterfrom
boergeson:fix/korean-burst-wpm
Open

fix(test): set burst start at composition begin for IME input (@boergeson)#7760
boergeson wants to merge 1 commit intomonkeytypegame:masterfrom
boergeson:fix/korean-burst-wpm

Conversation

@boergeson
Copy link
Copy Markdown

@boergeson boergeson commented Apr 2, 2026

Summary

  • Fixes Korean (and other IME) burst WPM showing absurd values (72,000+) in the heatmap
  • Root cause: setBurstStart was called at compositionend time, not compositionstart, making the time delta nearly zero and producing huge WPM from the division
  • Set burst start at compositionstart when beginning a new word so the timer reflects actual typing start
  • Skip overwriting burst start during compositionend in onInsertText
  • Guard against zero/negative time deltas in calculateBurst

Closes #7300

Test plan

  • Type in Korean (or another IME language) in words/time mode
  • Check burst heatmap in word history -- values should be reasonable, not 72,000+
  • Verify non-IME typing still records burst correctly
  • Verify NaN results no longer appear for single-character Korean words

…typegame#7300)

Korean IME burst WPM showed absurd values (72,000+) because
setBurstStart was called at compositionend, not compositionstart.
The near-zero time delta caused division to produce huge numbers.
@monkeytypegeorge monkeytypegeorge added the frontend User interface or web stuff label Apr 2, 2026
@boergeson boergeson changed the title fix(test): set burst start at composition begin for IME input fix(test): set burst start at composition begin for IME input (@boergeson) Apr 2, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

frontend User interface or web stuff

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Korean language heatmap WPM calculation inaccurate

2 participants