fix #311788: gap between staff and end barline with courtesy clef

[MarcSabatella]

Resolves: https://musescore.org/en/node/311788

A recent commit fixed some issues with courtesy clefs,
but introduced a regression - courtesy clefs on invisible staves
were not being ignored, causing the segment to be marked visible
and rhen in turn the measure width to be recalculated
as if the clef were visible.
This puts the barline too far to the right,
with gap between the end of the staff and the barline.
The clef itself was still ignored in the width calculation,
so the discrepancy is not large -
only the size of the margins before and after the clef.

Fix is to skip invisible staves in the code that is checking
for measures that need their width recalculated.

AIn debugging this, I became aware this clef on an invisible staff
was also appearing to affect the skyline of the visible staff,
leading me on a wild goose chase,
But that turns out to be an illusion -
the clef is not added to the skyline of the visible staff at all.
It's just that the skyline drawing code
(used when turning "Show Skylines" in the Debug menu)
is also failing to skip invisible staves.
So I have fixed that as well.

[AntonioBL]

Thank you very much Marc for the investigation.
Indeed, I had the same spacing problem when courtesy clefs were hidden, that's why I introduced the measure width recalculating step. However, I didn't take into account the fact that courtesy clef segments in invisible staves do not necessarily return visible=false.

Maybe a possible different fix is for the segment to return false when queried as segment->visible() if it contains only elements belonging to invisible staves. But maybe this would break other things in the code.

[MarcSabatella]

Actually, Segment::visible() does already do that, and that's why the initial assignment of "wasVisible" in the code here was "false". The problem is that the subsequent calculations are not taking staff visibility into account, so we think there is a clef to show, and thus we set visibleInt to 2. And then a couple of lines later, that results in the call to setVisible(true). Thus, the later comparison of "wasVisible != clefSeg->visible()" is true, and that's why we are calling computeMinWidth() when we shouldn't.

[AntonioBL]

Ah, ok. So completely jumping that part of code if the staff is invisible, as you did, is probably the only way to solve it.
Thank you. 😃

[igorkorsukov]

@MarcSabatella Could you please port the changes to the master