stems !

• Oct 4, 2012 - 22:02

herebelow copy of my post in the usual forum

any file seems to be affected
Par robert leleu, le octobre 4, 2012 - 9:57pm.

could you give a link to the technology preview forum

atached is a stem affected file (by the last linux nightly), which opens nice with the sept 13 nightly

Attachment Size
C’était_dedans_un_petit_bois.mscz 4.1 KB


In reply to by [DELETED] 5

I can't output a PDF on my Windows machine but here's a screenshot of the example file opened in revision c47cad0 on Windows 7 64 bit


It's not as badly affected as some of my files but I don't have anything free of copyright which I can upload at the moment.

It does share characteristics with every file I've seen which may or may not be relevant. The centre of the page is affected the most, there is some effect on the right but the left is almost entirely problem free (in this case it is free of problem stems)

I forgot to mention that I did not see this problem with revision 8d79909 from Sept 28th or anything earlier (I can't remember if I used any of the versions between then and c47cad0

Attachment Size
using_c47cad0.PNG 57.08 KB

In reply to by [DELETED] 5

Sorry, I've realised there was a mistake in my post yesterday.

I've just installed half a dozen previous nightly builds and found the problem first occurs in revision 96d2cb0. Earlier I said that 8d79909 didn't exhibit the problem but I must have downloaded but not installed that version as I can now see that it does have the problem stems.

So it seems the problem is caused by changes between revision 2012-09-23-ba582c2 and revision 2012-09-26-96d2cb0.

At the moment I don't have a build environment so haven't checked if there were commits between those two revisions which were not provided as binaries as I wouldn't be able to test them.

Edit: I was going to remove the incorrect information from the bottom of the post with the inline image to avoid confusion but I've realised the (12 hour?) edit window has expired.

The bug is a floating point rounding problem. Its a bad idea to compare floating point numbers in a naive way like integers. Affected are the first and last stem of beamed note groups. As this problem is platform dependent and hard to reproduce (i did not see it on linux) please check again with latest version.

In reply to by [DELETED] 3

I don't normally use Linux but thought I'd have a look at this out of curiosity.

I do see the problem on Linux using revision deb8c96 which is the latest build available (I installed Linux to test this but haven't gone as far as building the latest code).

Even the sample score which opens when you open MuseScore immediately displays the problem.

The test environment was Ubuntu 12.04.1 desktop 32 bit installed with default installation settings then updated but nothing additional installed afterwards except 2 MuseScore dependencies which were not in the Ubuntu packages (Qt-WebKit Library and Qt-Designer Library) and VMWare Tools as I used a virtual machine.

Ah sorry there was a typo I meant that not always the first stem is affected. In the pic above you see that only the measures in the middle have this problem.

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