Flag of crossed voice can overlap note in other voice

The flag should not overlap the notes, I agree.

However, the other differences between MuseScore rendering and the "expected result" are more personal taste. The fact that MuseScore does not by default beam three eighths togrther is in accordance with one set of standards, although it is certainly true that some editors do choose to beam them. The fact that the example shows the overlapping voice 2 note *after* the voice 1 note is, I think, less standard, although doubtless not uncommon. FWIW, we use Elaine Gould's "Behind Bars" as our go-to reference on such matters, although we also consult other sources. But Gould advises the downstem note go to the left when parts overlap as they do here, to allow the noteheads to overlap and save space. She doesn't show a case with flags, though.

It would be good to find some other source for this.

BTW, see http://musescore.org/en/comment/reply/45791 for recent discussion of this topic.

I have looked at this in the code and it is easy to simply reverse the two notes if the downstem has a flag, although I haven't thought through all possible cases. But we already do the exact same thing if the downstem chord has dots but the upstem does not. This would be essentially the same thing.

Actually, though, it ends up being the wrong thing to do if the *upstem* note also has a flag. Or at least, we'd need to clear that too, and then we lose whatever advantage we gained by flipping the note in the first place.

So, still open to suggestions here - and hopefully, some references to study.

Right now I have this as my solution:

It works well, but I still welcome discussion before I submit this.

FWIW, Sibelius always places the downstem note to the right, flag or no flag, contrary to Gould's advice to put the downstem note to the left so the heads can overlap. And if the upstem note has a flag, it collides with the downstem note. Finale always puts the downstem note on the left, but doesn't offset them at all, so it is not clear which stem goes with which note, and flags definitely collide.

My version above beats both of them on all counts here, I think, which shouldn't be a motivator I guess, but is :-)

I've continued to work on this, looking at behavior when there are dots and flags together. There are some collisions of dots with flags in some cases that won't be handled well, but that's a different part of the code. Meanwhile, though, I do have the chords themselves adjusting well for flags, including combinations of flags with dots:

Unless there are objections, I will submit this, as it's clearly an improvement.

Here is LilyPond 2.18.2's output of the image in #4:

Thanks for posting that. Huh. Now I guess it makes me wonder if we are wrong for assuming that it's "wrong" for flags to intersect noteheads. But it sure looks bad to me. Also, odd that the rests are offset by different amounts between beats 3 & 4?

Not often that I can say I like our output better than LilyPond's...

I went ahead and submitted this PR:

https://github.com/musescore/MuseScore/pull/1710

I guess there could still be doubt whether my solution is really better than any others, but to me, it looks best. It comes at the cost of taking slightly more space than LilyPond, but that's what you get when you resolve collisions.

Fixed in 70f5820f08

Automatically closed -- issue fixed for 2 weeks with no activity.