Musescore closing down on pressing 'play' on old(ish) file

• Jul 26, 2018 - 13:56

OS: Windows , Arch.: x86_64, MuseScore version (32-bit): 2.3.1, revision: 0899e9d

Hi all,
my latest version of Mscore is working absolutely fine. Decided to pull a score out from the dusty corner of my computer and have a listen. It shut Musescore down. Tried again and pressed the 'Rewind' button rather than the Play button. Same result. To eliminate my computer from the mix, a friend downloaded Musescore to his (new) computer, then the file in question. It shut down Musescore in exactly the same annoying way.
Methinks it must be something in the file. I can't remember which format I am supposed to upload for someone to have a butchers.

Comments

• Jul 26, 2018 - 15:01

Can't be that old, it got last saved with 2.3.1.
But indeed crashes when starting playback in 2.3.1 and in 2.2.1.
Not in 2.0.3 or 2.1 though, so yet another 2.2 regression?

Crashes due to due a failed assertion:
Fatal: ASSERT: "!_endElement || _endElement->type() == Element::Type::MEASURE" in file ...\MuseScore\libmscore\spanner.cpp, line 771 (:0, )

Reply

• Jul 26, 2018 - 16:35

It doesn't take a lot but now I am confused. Am I supposed to be renaming the file without the dot and comma and re-attaching it. Or does the dot and comma signify that the one I sent earlier is the backup and therefore the one that you asked for? Sorry for taking up your time.

Reply

• Jul 27, 2018 - 17:30

Those ties look like what happens if you select a note to tie but the next note is not of the same pitch. This is normal, the feature exists to facilitate notation of arpeggiate-and-hold figures. The octave jumps look like something that may have resulted from a bug in transposition that was fixed years ago, but older scores created before that fix would not automatically get fixed.

Reply

• Jul 27, 2018 - 07:30

OK, that file got created (at least last saved) with MuseScore 2.0.1

The code for that bogus volta:

      <Measure number="12">
        <Volta id="15">
          <endHook>1</endHook>
          <lid>288</lid>
          <anchor>0</anchor>
          <Segment>
            <subtype>0</subtype>
            <off2 x="5.58264" y="0"/>
            <offset x="-3.38342" y="-0.338342"/>
            </Segment>
          <beginText>
            <text>1.</text>
            </beginText>
          <endings>1</endings>
          </Volta>
...
        <Measure number="11">
          <irregular/>
          <Volta id="551">
            <endHook>1</endHook>
            <lid>288</lid>
            <anchor>0</anchor>
            <beginText>
              <text>1.</text>
              </beginText>
            <endings>1</endings>
            </Volta>

Reply

• Jul 26, 2018 - 17:34

The crash seems to be triggered by a volta from measures 11-12 that, instead of being attached to the measures themselves, is attached to notes within it. That isn't supposed to be possible, but somehow it happened, probably due to a bug in the version you originally created this with. Would be nice to figure out how that happened so we can see if it's still possible to create a file that is corrupt in this way. We should also look into how to automatically fix this problem in existing scores.

Meanwhile, you can fix the file by removing that volta and re-adding it.

Reply

• Jul 26, 2018 - 22:54

Thanks Marc. As you so correctly worked out, the volta is fixed and it now plays. However, that doesn't explain why there are those unexplained 8ve jumps (in either direction) and the ties disappearing into the wilderness.
I'm currently going through and deleting the ties, dropping/raising the 8ve and then re-tieing. A pain but at least its recoverable.
I won't ask how you worked out it was the volta that was the problem as I am certain it's way beyond my level of understanding. Thanks again.

Reply

• Jul 27, 2018 - 17:26

FWIW, 2014 was before the release of 2.0. So if this file was created originally using an experimental nightly build, I think we can put this to rest as "that's just what happens when you try using experimental nightly builds for real work". If it was created originally using 1.3, that's another matter. Could still be a bug in 1.3 that was fixed long ago. I still can't say what specifically about the way the volta is represented in the file is triggering the problem, though.

Reply

Do you still have an unanswered question? Please log in first to post your question.