Re: bar checks and independent time signatures

[Brian Barker]

At 15:57 22/10/2015 -0400, Michael Collins wrote:
> A couple days ago, I described a problem with a piece for two staves 
> in independent time signatures. After consulting this post, I moved 
> the timing translator and bar line engraver out of the score context 
> and into the staff context. While the midi compiles properly, the 
> music in the pdf runs off the page.
>
> Lilypond says that the bar checks in the top voice are failing. I 
> suspect that this is the root of the problem. I've run versions of 
> the code with each of the two voices commented out. When the voices 
> play separately, the pdf output stays on the page. Something about 
> combining the two voices is problematic.
>
> I'm genuinely at a loss, since moving the timing translator to the 
> staff context should make the voices completely independent. Any advice?

With a system containing two staves, Lilypond can create system 
breaks only where bar lines occur in both staves simultaneously. 
Music runs off the page when it cannot create appropriate system breaks.

The first bar in your lower staff contains the equivalent of four 
crotchets (quarter notes), the second 4/2 = 2, the third 4/3 = 1.33 
(approx.), and so on. The length of the music up to each bar line in 
the lower staff in crotchet units (rounded) is thus:
4
6
7.33
8.33
9.13
9.8
10.37
10.87
11.32
11.72
12.08
12.41
12.72
13.01
13.27
13.52

The first bar line, at four crotchets, coincides with a bar line in 
the upper staff, so a system break can occur there - and does. The 
second, at six crotchets, coincides with the middle of the second bar 
in the upper staff, so you could engineer a system break there by 
putting an invisible bar line at that point in the upper voice. But 
nowhere else do the ends of any *notes* - not just bars - coincide in 
the two voices. So you simply cannot have a system break without some 
note in one or other voice itself being broken - which doesn't happen 
in musical notation, of course. In principle you could divide a note 
and tie it across the break, but there would be no way to represent 
the strange fractions of note length you would require, would there?

The music runs off the page only because of the impossibility of 
(later) system breaks. You can see the full engraving if you either 
reduce the staff size or increase the paper width. You will also then 
easily see that the upper staff extends beyond the lower one, as you 
can see from the values above: the lower voice lasts approximately 
13.52 crotchets but the upper voice sixteen. You need two-and-a-bit 
crotchets' worth of rest at the end of the lower staff to make 
everything fit. You'd need a fearsome time signature to achieve this 
- which is left as an exercise for the reader!

Brian Barker