Description: Selection scene keys churn when selection is effectively unchanged, causing unnecessary pass=selection invalidations and repaint work. Make key inputs deterministic and geometry-only where possible; remove unstable/transient fields.
Acceptance: Idle/hover with unchanged selection does not continuously invalidate selection tiles.
Description: Selection scene keys churn when selection is effectively unchanged, causing unnecessary `pass=selection` invalidations and repaint work. Make key inputs deterministic and geometry-only where possible; remove unstable/transient fields.
Acceptance: Idle/hover with unchanged selection does not continuously invalidate selection tiles.
Probably moot - needs a decision rather than the work. Checked 2026-09-05 during the roadmap audit.
This was written against the V2 tiled renderer's dedicated selection pass, where a selection tile cache was invalidated per scene-key change. That architecture is gone:
kV2UseSelectionTileCache = false (lib/constants.dart:329), so the selection tile cache does not run at all
compositing has since moved to the unified z-axis and band model (scene_z_order.dart, sceneBands, qt_tiles.dart), where selection is not a separate cached pass invalidated on its own key
So "pass=selection invalidations churn on idle/hover" describes a code path that no longer executes. The acceptance criterion cannot really be tested as stated.
Not closing it unilaterally, because the underlying concern - scene keys churning on effectively-unchanged state and causing needless repaint work - could still be true of the CURRENT band/tile keys, and nobody has measured that. But if it is, it wants re-filing against the band compositor with fresh evidence rather than staying open on the old wording.
Suggest either closing this or rewriting it against qt_tiles.dart after an actual idle-hover probe.
Probably moot - needs a decision rather than the work. Checked 2026-09-05 during the roadmap audit.
This was written against the V2 tiled renderer's dedicated selection pass, where a selection tile cache was invalidated per scene-key change. That architecture is gone:
- `kV2UseSelectionTileCache = false` (`lib/constants.dart:329`), so the selection tile cache does not run at all
- - compositing has since moved to the unified z-axis and band model (`scene_z_order.dart`, `sceneBands`, `qt_tiles.dart`), where selection is not a separate cached pass invalidated on its own key
So "pass=selection invalidations churn on idle/hover" describes a code path that no longer executes. The acceptance criterion cannot really be tested as stated.
Not closing it unilaterally, because the underlying concern - scene keys churning on effectively-unchanged state and causing needless repaint work - could still be true of the CURRENT band/tile keys, and nobody has measured that. But if it is, it wants re-filing against the band compositor with fresh evidence rather than staying open on the old wording.
Suggest either closing this or rewriting it against `qt_tiles.dart` after an actual idle-hover probe.
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Description: Selection scene keys churn when selection is effectively unchanged, causing unnecessary
pass=selectioninvalidations and repaint work. Make key inputs deterministic and geometry-only where possible; remove unstable/transient fields.Acceptance: Idle/hover with unchanged selection does not continuously invalidate selection tiles.
Probably moot - needs a decision rather than the work. Checked 2026-09-05 during the roadmap audit.
This was written against the V2 tiled renderer's dedicated selection pass, where a selection tile cache was invalidated per scene-key change. That architecture is gone:
kV2UseSelectionTileCache = false(lib/constants.dart:329), so the selection tile cache does not run at allscene_z_order.dart,sceneBands,qt_tiles.dart), where selection is not a separate cached pass invalidated on its own keySo "pass=selection invalidations churn on idle/hover" describes a code path that no longer executes. The acceptance criterion cannot really be tested as stated.
Not closing it unilaterally, because the underlying concern - scene keys churning on effectively-unchanged state and causing needless repaint work - could still be true of the CURRENT band/tile keys, and nobody has measured that. But if it is, it wants re-filing against the band compositor with fresh evidence rather than staying open on the old wording.
Suggest either closing this or rewriting it against
qt_tiles.dartafter an actual idle-hover probe.