file corruption (solved)
I had a corrupted file and about no hope not to loose all of my work. Nevertheless I unpacked the mscz-file and looked into it with a text editor. It was obvious that something went wrong with the time signature. The original sheet got a 3/4 pick up in an 4/4 time signature. But the file showed a 3/4 time signature over the whole sheet. It had been easy to correct that and afterwards the file was loaded smoothly by musescore.
I never had such a file corruption in the last 2 years with musescore and I can't imagine that it has been my fault this time . But let me come to my question. Is there a description of the inner format in which musescore files are stored? Maybe it could help to fix a file corruption in case it is not as obvious as in this special case.
Comments
I'm glad you were able to sort it out! The file format isn't documented, in part because details may change from release to release and we don't want people relying on things that may end up changing. But if you are sufficiently motivated and experienced, you can browse the source code on GitHub (see Development link in menu at right).
FWIW, the type of corruption you describe can happen if a time signature was deleted using an older version of MuseScore (1.3 or earlier). This wasn't supposed to be possible, but those earlier versions would occasionally allow it and then not detect the resulting corruption. Current versions do detect those errors introduced by older versions. Certainly current versions may also have their own problems, but your description sounds suspiciously like what I described. So it's possible whatever happened was something like that. If you still have the uncorrected file, we'd love to examine it to try to understand. Because you're right - corruption is never the user's fault; it always indicates a problem with MuseScore.
Marc, thank you for your attempt of an explanation. No other version than 2.0.3 had been involved in the editing process. I only remember that after the last successfull save I had only done some minor changes or formating.
Indeed I made a copy of the file before trying to fix the corruption. It had been work in progress and the bass clef line had not been filled yet.
In reply to Marc, thank you for your by SlyDr
Hmm, this is strange. It looks like somehow the 4/4 time signature in bar 1 got changed to 3/4. After all, there really shouldn't be a 3/4 in bar 1 - it should have been 4/4, but with the actual duration changed to 3/4 in Measure Properties. And I assume that's what it looked liked originally.
I can't say I've seen anything quite like this before. Anything you can remember about what you might have done to trigger the change would be helpful. Again, it's clearly a bug - even if you deliberately changed the time signature, it should have updated everything and not caused a corruption. But we need to be able to reproduce the problem in order to figure out how to fix it.
Sadly I can't exactly tell what I had been doing with the score. I only remember doing some minor changes and wanted to continue with the bass clef the other day when I noticed that the file was corrupted.
I regret that I can't be of any help regarding the cause of the corruption.
I read the advice somewhere in this forum not do trust in automatically stored backups too much.
I did not account for this advice but had been lucky this time. But I think I have learned my lesson and hopefully I will be as careful with my scores as I use to be with my word documents which tend to get corrupted by chance, too.
In reply to Sadly I can't exactly tell by SlyDr
"No other version than 2.0.3 had been involved in the editing process."
I think it's the main point.
And I will bet that is/was this issue, involving a confusion (or clumsiness) about how a pickup measure may be showed. Issue fixed for the 2.1
#152506: Introduce a time signature in a measure which the Actual duration is exactly identical leads to corruption/crash
Indeed, I can reproduce exactly the same corruption, with the 2.0.3
1) New default score (so, 4/4 time sig)
2) In measure 2: change the actual measure duration to 3/4, and drag and drop in the same measure, the 3/4 time signature. From this moment, the bug is en place.
3) Enter some notes
4) Select the first measure (with 4/4 time sig) -> Ctrl + Del
5) Save/Close/Reload
Result: corruption identical to the initial attached file.
pickup.mscz
EDIT: so, the advice is to update to the 2.1 (if not already done)