Comment #2 on issue 3145 by [email protected]: flaws in accidentalGrouping algorithm
http://code.google.com/p/lilypond/issues/detail?id=3145
In both the third and fourth chord, the g sharp in the lower voice is more distant to the note head than the c sharp. It should be exactly the opposite. Somehow it seems that the `accidental sort' doesn't restart properly, this is, the topmost accent in a chord should be nearest to the note head if possible – and here it definitely *is* possible.
Additionally, in the third chord, it looks as if the g sharp optically belongs to the upper voice. Given the good skyline stuff you've developed, I see no reason why this is so.
