[Musicxml Export] - Missing spanners caused by spanner stop not in the first track

• Oct 7, 2019 - 10:50
Reported version
3.2
Priority
P1 - High
Type
Functional
Frequency
Many
Severity
S1 - Blocker
Reproducibility
Always
Status
closed
Regression
No
Workaround
No
Project

Hello,
I just finished transcribing this piece. Since I can't reproduce the problem snippet myself, I have to attach all the score. Generally, I found lots of hairpins missing when exporting to Musicxml. For example, (below I count from the first complete measure), m8-9, there are lots of missing hairpins when using the fixed Musescore 3 file, including some woodwinds, horns and strings. When I use the old Musescore 2 file, the hairpins are there. I export both scores using Musescore 3.3 rc, and the problem is the same. I checked the Musescore file code, and both files don't hide such hairpins in these measures. I suspect all braille scores I have published may be effected by this strange problem, and for this work, I have to check the braille bar by bar using the old file. What's wrong with the export? Thank you in advance!

Regards
Haipeng


Comments

Priority P1 - High

I can confirm the missing hairpins after opening the MSCZ file, exporting to MusicXML, then reimporting that. Specifically, I checked the clarinet hairpins in bars 8-9 as notated (9-10 as per the internal measure numbering), I'm would bet whatever others are missing, the cause it the same.

I just tried again, and its strange that exporting as parts doesn't generate this problem. But now I found this problem happens in some scores, but not in others. For example, I want to transcribe this into braille:
https://musescore.com/user/2749876/scores/5672184
But just now I found all hairpins after part 5 disappeared in Musicxml. I went back to Wieland score, and found all hairpins disappear from part 4 on. However, all hairpins are exported in Beethoven's 9th symphony. So it's a puzzle for me, and I don't know when it happens. Now I have give up all Musescore-related braille transcription in case I'll do harm to blind musicians. The only solution is, I have to export all staves into parts and compare them one by one, bar by bar, with the braille full score--What a terrible work if there are 40-50 parts! I'm wondering why the hairpins in lower parts can't be exported.

Haipeng

Oh, don't close, this is still the issue. I found other scores without irregular bars having this problem. I just transcribed Dvorak's cello concerto and published on my site. Big failure, and misleading to blidnd musicians!

Haipeng

Ok, here's another example chopped to the shortest I can do. The 3 versions contain the same music with 2 or 4 bars deleted, and The deletion of some bars will make hairpins of some parts appear, at last to fully exported in 20 parts. I really can't find any reason for these strange behavior. The only reason I think is, Musescore can't export hairpins more than a certain amount. Am I right?

Haipeng

Sorry for too many posts today, but I'm really very anxious and frustrated with this issue, which almost makes me break down all works but simple piano and violin pieces.

It seems that I have found the culprit. Please see the attached clarinet part of Ravel. The hairpins in the cadenza section have their ending clearly stated in the original Musescore file, but are all lost in Musicxml export, which ends up with an exhausted use of the wedges. So there are two points:
1. Please solve the missing ending problem;
2. We should raise the exhausted levels to a higher amount, in case there are cases a staff has lots of special lines occupying more than 6 levels. I met a Sibelius file where the piano solo part has 12 levels of slurs because of its complicated cross-staff phrases, so I have to manually modify the Dolet plugin to export up to 24 levels of slurs, ties and wedges.

Haipeng

Attachment Size
culprit.mscz 40.83 KB
Title [Musicxml Export] - Strange Missing Hairpins [Musicxml Export] - Missing spanners caused by spanner stop not in the first track
Status active PR created

Pull request created, see: https://github.com/musescore/MuseScore/pull/5423.

Root cause is that spanners (hairpins, ottavas, pedals and textiles) are not stopped when the stop is associated with a note in track 2, 3 or 4. This results in the list of currently open spanners in the export filling up completely, meaning no more spanners can be exported.

Please note that increasing the maximum number of open spanners leads to invalid MusicXML, so I am somewhat reluctant to do that. See http://usermanuals.musicxml.com/MusicXML/MusicXML.htm#ST-MusicXML-numbe…. Suggest you open a separate issue for that, where we collect the relevant info. I would like to know how this affects interoperability, but can only test Finale Notepad.

Thank you very much! I'll raise this request of limitation increase on W3C Music Notation Community Group's Musicxml 3.2 issues.

That would indeed be the correct approach to get this issue addressed. If the maximum numbers are increased in MusicXML, it is trivial to update MuseScore.

Fix version
3.3.1