Close/reload crash on XP

• Jul 26, 2010 - 23:39
Type
Functional
Severity
S2 - Critical
Status
closed
Project

Musescore r3235 and subsequent crashes on XP with reload/close for certain test scores.


Comments

In my test of the "LockAndKey" score mentioned above with r3235 on XP, deleting measures 31 to the end will result in a portion that closes properly. Deleting 32 to the end results in a portion that crashes. Inserting an empty measure before measure 31 and deleting after the empty measure to the end also crashes, so the crashing doesn't appear to be due to some funny business in the original measure 31. So the crashing appears to be solely dependent on the length of the score.

Here's a really easy way to demonstrate this bug. On XP open 0.9.6.1 that has been set to factory defaults. Insert 60 measures before the first measure in the demo score Bilder einer Ausstellung. Then File->Close. Musescore crashes.

Status (old) active fixed

Fixed in r3342 and r3344. We will issue a MuseScore 0.9.6.2 to address this particular issue and couple of others.

Where would I be able to find these? I look under "nightly" on the prerelease link, but the latest posted was r3325.

This prerelease (0.9.6.2) seems to close properly on my system, regardless of which files I load or close.
I had the problem described in this thread, but no longer in this prerelease.

Thank you for fast and excellent work!

Almost, but no cigar. With 0.9.6.2, the example of 60 bars ahead of the demo seems to reload and close fine, but the LockandKey mentioned above sometimes reloads, sometimes closes, but often still crashes on close or reload, at least on my XP. I've spent the last few minutes searching for a pattern, but I haven't found it yet ...

The prerelease 0.9.6.2 seems to have done the trick for me. I can open, close, reload, etc. with no crashes.

I'm sure you know, but FYI the splash screen still shows 0.9.6.1.

Ach, REdwards is correct indeed. I checked again, more thorough after reading his post, and the problem seems to persist if two or more long scores are closed manually (Ctrl+w) very quickly.
The problem does not appear for me when closing the entire program via the X button.
When watching the memory usage via Task Manager, it is evident that the memory mscore.exe increases rapidly prior to the crash.

I'm sorry that your work did not completely succeed, but I salute your stamina and idealism in this work.

I've tested 0.9.6.2 with a number of long scores. Each causes an XP crash on reload. What information can we give to developers to help track down this problem?

How to consistently crash Musescore 0.9.6.2 on XP: Make the LockandKey test score (#1 above) twice as long by appending a duplicate of the original and save it. Open the resultant score and raise the first note by a half step. Then File->Close (discard changes). On my machine, Musescore will always crash.

I took another look to the crash. I still don't know why Windows XP reacts differently on delete objects.
The last remaining crash was due to deleting tuplets in the Measure destructor. Not sure why ... I will do another prerelease asap.

Pass out the cigars. This one looks like a winner. It won't crash for me during reload or close for all the scores that had problems previously.

Just now checked the second prerelease, doesn't work ( stuck on startup ) . Back to first prerelease, doesn't work too! But the first time I checked prerelease #1 it worked!?!?

Fixed?

frareinif, this thread is about a crash when closing a score. If you are experiencing a crash at startup see revert to factory settings . For further help please use the forums and include all the necessary information for others to help you (see [[nodetitle:How to report bug or ask for support]] ).

1. The first time I reported my issue it has been merged with this thread.
2. Revert to factory settings didn't help, the app is stuck in the startup screen causing a permanent 50% cpu load.

@frareinif It could have been merged by mistake or wrong assumption it was reporting the same bug. Could you point back to your original bug report?