MuseScore
MuseScore copied to clipboard
musescore
[MEGA THREAD] Tracking issue for known kinds of corruptions
There are several reports of score corruptions, and I think it would be good to list them in one place and categorize them. So here we go:
- [ ] https://github.com/musescore/MuseScore/issues/15185
- https://github.com/musescore/MuseScore/issues/15360
Crash Replication Test Site.mscz - https://musescore.org/en/node/341414
Dum av dig.mscz - https://musescore.org/en/node/341424
Gubben i lådan.mscz
- https://github.com/musescore/MuseScore/issues/15360
- [ ] Corruption in main score; some notes in some measures too long; seems related to mysterious
<location>tag??- https://github.com/musescore/MuseScore/issues/15519
Clemens 4-handed.mscz - https://musescore.org/en/node/341430
Crushdance.mscz - https://musescore.org/en/node/341127#comment-1163573
The_Physical_Impossibility_bug_report_samples.mscz - https://musescore.org/en/node/340301 https://musescore.org/en/node/340353
Mechanically Separated Chicken.mscz - https://musescore.org/en/node/340301#comment-1159814
Piece for Cello 1.mscz - https://musescore.org/en/node/340552
StringQuartet1.mscz - https://musescore.org/en/node/340126
866.mscz - https://musescore.org/en/node/339824
A Nightingale Sang in Berkeley Square_2.mscz - https://musescore.org/en/node/339686
Piel Canela_Bbmaj__ls04.MS4_.mscz - Seemingly https://musescore.org/en/node/339315 too, but that one should be taken with a grain of salt because it's from 1.x
- https://musescore.org/en/node/338567
Momentum_-_Cosmos_G..mscz - maybe this one too: https://musescore.org/en/node/339165
Missa_Rearranged V4.mscz(it looks like the corruption is in the parts, but then breaks the main score on read; the parts corruption looks like it's caused by the first corruption mentioned) - https://github.com/musescore/MuseScore/issues/15689
Polyamateur 4 MS4.mscz.zip
- https://github.com/musescore/MuseScore/issues/15519
- [ ] Corruption in main score; tuplets seem somehow lost
- https://github.com/musescore/MuseScore/issues/15723
The Merry-Go-Round of Life – Joe Hisaishi (FULL_ Orchestral Transcription).mscz
- https://github.com/musescore/MuseScore/issues/15723
- [ ] Sort of corruption in main score due to actions with measure repeat: https://github.com/musescore/MuseScore/issues/15844#issuecomment-1382866721
- steps:
- Create score with measure containing notes, another measure containing measure repeat
- Copy contents from measure containing notes to measure repeat measure
- To see the corruption, press Delete on the measure that previously had the measure repeat: you see a strange empty measure now
- steps:
- [ ] #15490 which is currently in need of info
- [x] https://github.com/musescore/MuseScore/issues/15679
- [ ] Zip file corrupted -> no recovery possible
- https://github.com/musescore/MuseScore/issues/15113
Bossa_Lemanica.mscz - https://discord.com/channels/818804595450445834/818804809784229898/1057744302765256874 (I know discord links are useless but it would be useless anyway, since a file full of nulls is not very interesting)
- https://musescore.org/en/node/342465 file became 0KB
- https://github.com/musescore/MuseScore/issues/15113
- [ ] A very peculiar case where the main mscx file is missing from the mscz file but the rest is there
https://musescore.org/en/node/340829#comment-1163670
.Orquestrado.mscz(Oh wait, maybe this is because this person had the great idea of starting the filename with a ".") - [ ] Another very peculiar case where the main mscx file is not at root level in the mscz file, but in "Users/perfc/Downloads" inside the mscz file
https://musescore.org/en/node/340775
Carol of the Bells.mscz
If you find any corrupt score in MS4 or a new report thereof, please comment here, and I'll add it to the list!
After one evening of trawling through the musescore.org issue tracker, it has become obvious that the <location>-related (?) one is by far the biggest. From looking at the scores, nothing becomes immediately clear, except that in a lot of cases a <location> tag is present around the location of the corruption.
However, I'm starting to suspect that the cause of these corruptions is in fact the same as the first corruption in the list. Initially, the main score would still be intact, and only the parts would be broken; but after some saves and especially re-opens, the corruptions propagate to the main score. So far, this is supported by the observation that in one case, the score opened without any corruptions after I had extracted the main mscx file outside of the mscz, effectively deleting the part scores. I'll try this with more scores.
I know @vpereverzev thinks that the majority, if not all these issues are fixed. Can you confirm Casper? Might make sense to actually just assign incomplete issues here individually and remove the megathread.
I also suspect that the cause of the first two points is fixed, but it is difficult to confirm that this was indeed the cause in all those cases. Maybe the best way is to wait a while and watch the forums to see if indeed no new corruptions are reported.
But yeah, for the other cases, there doesn't really seem to be a pattern, and they have been reported only once, so it makes sense to handle them individually. I don't think they are new in MS4 either, actually.
(Except the last two; those are caused by the new file format (edit: actually not so sure about that even), and may be related in a way.)
By the way, here is a mega mega thread about those mysterious ZIP file corruptions (where the file either becomes 0 bytes, or becomes filled with only zeros and is therefore useless): https://musescore.org/en/node/271185
So that's really not new... but also not fixed by the changes to the ZIP writer in MS4.
Good news: since 4.0.1 I have seen no reports anymore of new corrupted files. I think this is statistically significant, so that we can conclude that the first two big ones can be marked as fixed.
Bad news: there are two new ones (I'll add them to the issue description)
- for some people, spanners (hairpins, voltas, lines.....) are missing in part scores
- https://musescore.org/en/node/341765
- https://musescore.org/en/node/343502
- there's this corrupted case: https://musescore.org/en/node/343454 (the score is not detected/reported as corrupted, but the log output clearly shows that it is) (it's a single one so far, but quite a strange one, since it looks like staff indices somehow got out of sync between the master score and the part score)
I've now created distinct issues for the remaining ones, and labeled them with a new corruption label to keep track of them. So I think we're ready to close this thread. When a new corruption is found, users can create a separate issue for it, and if it turns out to be caused by the same thing as an existing report, we can just link them together and close one of them, like we always do.
