Hangul renders as decomposed jamo in editor on Linux/Qt 6.11, but copied text remains correct

Environment

  • Anki 26.09.2 (bb0dd6d1)
  • Python 3.13.15
  • Qt 6.11.2
  • PyQt 6.11.0
  • Fedora 44, kernel 7.2.9-201.fc44.x86_64

In Anki’s Add/editor field, some Korean Hangul sequences are rendered incorrectly as decomposed jamo even though the underlying text remains correct.

Minimal reproduction:

  1. Open Add.
  2. Paste 가나다.
  3. 가 renders normally, while 나다 appears visually decomposed.

Other tests:

  • 가, 나, 나다, 다라, 라마, 바사 display correctly.
  • 가나다 breaks after the first syllable.
  • 안녕하세요, 난방보다 싸요, and 저는 오늘 집에 있어요 display correctly.
  • Copying the visually broken 가나다 back out of Anki pastes elsewhere as normal 가나다, so the stored text appears to be correct and this seems to be a rendering/shaping issue.
  • Korean displays normally in Firefox/Google Translate.
  • Restarting Anki does not help.
  • QTWEBENGINE_CHROMIUM_FLAGS="--disable-gpu" does not help.
  • QT_IM_MODULE=ibus IBUS_USE_PORTAL=1 anki does not help.

Screenshot attached: