Layout Invalidation Performance

[thecoolwinter]

Description

• Added a new debug mode for visualizing line fragment invalidation.
• Adjusted the layout pass to avoid typesetting lines that don't need layout.
  • Made a distinction between 'forced layout' via setNeedsLayout and 'continued' layout where a line previously scanned in the layout pass was updated.
  • Due to that, I was able to check if a line fragment actually needed typesetting or just potentially needed to have it's position adjusted.
  • Added a new method to update a line's view's positions during layout.
• Removed the unnecessary lineRange variable on the LineFragment class.
• Adjusted the use of documentRange on the LineFragment class. It's now updated during layout, simplifying various methods.

Related Issues

• N/A

Checklist

• I read and understood the contributing guide as well as the code of conduct
• The issues this PR addresses are related to each other
• My changes generate no new warnings
• My code builds and runs on my machine
• My changes are all related to the related issue above
• I documented my code

Screenshots

Before:

Screen.Recording.2025-07-21.at.4.21.43.PM.mov

After:

Screen.Recording.2025-07-21.at.4.17.29.PM.mov

[thecoolwinter]

Posting this a little after the fact, but here's a profile from a regular editing session before this change. Each keystroke the profiler marks as a hang, and over the duration of the editing the minimap and textview take up a combined 36% of CPU time

After these changes, theres very very few detected hangs* and text layout now only takes up 5% of CPU time that's an 87% reduction

• (all under 60ms and related to tree-sitter)