If last staff is shorter in duration than other staves, music beyond that duration is lost from other staves

[mjwalter]

A test case is in my forked repository as TC-blank2.xml.

Staves with empty <layer> elements render fine. However, if such a staff is the last staff in the system, all staves appear blank.

The test case demonstrates this behaviour: the first measure has an extra (fifth) staff, and the parts render well, but in the second measure, the empty <layer> is the last staff, and therefore triggers the bug. If this layer is filled, the problem disappears.

(I would appreciate if you could point me to the problematic section of the code so I can try my hand at it myself!)

[zolaemil]

not sure, why... if you think it's a meitovexflow issue, I suggest debugging it with a m2v test case. Perhaps a bug in the selection of staves when slicing (meilib slicing methods)? you can find tests for that in test_meilib.html (actual tests implemented in runtest_meilib.js)

[mjwalter]

Can you think of anything which relies on the last staff only for information? I'm thinking that the duration of the whole measure is calculated by looking at the last staff.

[zolaemil]

Voices must have the same duration in order to achieve correct alignment.

[mjwalter]

But an empty layer shouldn't wipe out the material in the other staves. The music renders perfectly well if the blank staff is not the last. Instead of basing the duration of a measure on the values found in the last staff, it would be better to look at the duration of the longest staff.