Selection latency / slow in large scale project (compared with MS3)

Issue type

UX/Interaction bug (incorrect behaviour)

Description with steps to reproduce

Medium to large scale projects suffer from this significantly.
Did some search and found #9174 marked as close, but I think it hasn't been completely fixed yet, only somehow mitigated - when comparing to early MS4 releases.

P.S. maybe a new issue type for performance issues like this?

Supporting files, videos and screenshots

Here's a test file created in 3.6.2 (renamed to .zip). It's just 6 staffs filled with 16th notes for about 100 measures (~10k notes):

Stress Test.zip

Here's selection speed in 4.5.0 release:

And in 3.6.2:

... or simply, just click a measure.

What is the latest version of MuseScore Studio where this issue is present?

4.5.0 Release

Regression

Yes, this used to work in MuseScore 3.x and now is broken

Operating system

Win11 24H2

Additional context

It's significantly faster in an empty project - but so is MS3, and there is still a perceivable difference. Discrepancy becomes more obvious with larger projects.

I was suspecting an audio engine issue, but in an empty project the sound actually comes before selection highlight (with MS basic). And changing buffer size seems having no effect on this.

Less important info:

These two gifs were representing my "real life" editing speed. I was always waiting MS4 to "complete" the selection before doing anything else, worrying about triggering bugs or freezes - but after this experiment, it seems that it's my overthinking. I actually might just act as fast as I want, then wait for MS to respond (remind me of early Android days). Will try to see if this would cause problem...

Checklist

• This report follows the guidelines for reporting bugs and issues
• I have verified that this issue has not been logged before, by searching the issue tracker for similar issues
• I have attached all requested files and information to this report
• I have attempted to identify the root problem as concisely as possible, and have used minimal reproducible examples where possible