fix #57646: corruption on paste with segment annotation in voice > 0 #1971
Conversation
| @@ -324,9 +324,10 @@ bool Score::pasteStaff(XmlReader& e, Segment* dst, int dstStaff) | |||
|
|
|||
| // be sure to paste the element in the destination track; | |||
| // setting track needs to be repeated, as it might have been overwritten by el->read() | |||
| el->setTrack(e.track()); | |||
| // preserve *voice* from source, though | |||
| el->setTrack(e.track() + el->voice()); | |||
There was a problem hiding this comment.
Is there a chance e.track() would already have a voice offset added in? It was always the first track of the staff in my tests. I suppose safer might be to convert e.track() to staffIdx and then back to track before adding the voice? The original code here did in fact use staffIdx. That was changed here:
There was a problem hiding this comment.
I don't get all the ins and outs just yet but it feels definitely better and more understandable to use dstStaffIdx * VOICES + voice instead of relying to track(). Even if it's set at the start of the loop to dstStaffIdx * VOICES
There was a problem hiding this comment.
I'm inclined to agree, but I figured the change was made for a reason. I could also see taking (e.track() / VOICES) * VOICES. @wschweer, is there a reason to prefer one of these?
|
See 210d567 |
This definitely fixes the bug, as well as another where the pasted annotation moves to voice 1. I don't think this breaks anything else, but it would be good to have someone else think through this (see my notes in the issue report) to be sure I am not missing something.