Rests when nominal and actual time signatures are different
When the nominal and actual time signatures in a bar differ, MuseScore treats a full measure rest as referring to the nominal time signature rather than the actual. This results in "File corrupted" warnings when re-opening a score, although it seems to have no effect on playback or, under most circumstances, score layout.
I decided, at one point in the creation of my last score, to fix each of the warnings by replacing the full measure rests with actual rests of the correct value, and discovered this isn't easy or obvious. In many instances, replacing, say, a whole measure rest with a real whole-note rest would result in a bar that contained only a whole measure rest at the beginning despite the length of the bar being--again, for example's sake--six quarters long. When the nominal and actual time signatures are the same, MuseScore adds the extra half-note rest, which is the expected and correct behaviour, but does not always or reliably do so when the time signature values are different.
Worse, I'd sometimes end up with a measure with, say, a sixteenth rest at the end when the correct value should have been dotted-eighth, but MuseScore wouldn't let me delete the rest (so it would auto-correct to a dotted-eigth), nor would it let me select the rest and click in the correct value. The only way to overcome this problem was to start at the beginning of the bar and convert every rest to the smallest value necessary to describe the bar (say, 17 thirty-second rests), then systematically reduce the thirty-second rests to sixteenths, then to eighths, then to quarters, etc. in order to get a bar with the correct rest count.
Comments
How were you getting full measure rests in the measures with different actual and nominal duration? Normally, changing actual duration should remove the measure rest automatically. Can you post precise step by step instructions to reprocuce this scenario?
Also the other part of what you describe - in general, when reporting problems, assume we always need sample scores and precise step by step instructions in order to see what is going on.
In reply to How were you getting full by Marc Sabatella
Marc --
It sometimes happens that a score becomes so complex that trying to reproduce a bug becomes impossible. I *definitely* had whole measure rests in non-"actual-length" bars in the score, Ch'ien, which I attached with another post. I know because a) I could see them, b) there were many, many reported errors of the type "Expected: 17/32, Found 6/8", c) I have callouses on my fingers from hours of removing the whole measure rests (and others that made no sense) and manually "re-resting" bars.
I cannot, however, reproduce the bug now that everything's been fixed in the original score, nor can I reproduce the bug with a simpler test score. Still, it's there. Possible culprits are my extensive use of cross-staff beaming, and the sheer amount of confusion to MuseScore caused by the quantity of irregular length bars.
I'll be engraving a second piece from the same work. It presents the same notational challenges as Ch'ien. Perhaps the problems I've described will show up, in which case I'll be able to give you steps to reproduce the bug(s).
As for the Ch'ien score, just re-opening it for bug-testing purposes reveals more inexplicable changes to my layout, per my other bug report. Luckily, I do have a good PDF, and the videos I've made of the score show the correct layout, but as far the MuseScore file itself is concerned, it's pretty useless now, given the errors (both consistent and random) introduced by opening it.
I'll see how things go when I build and test 2.0.1.