Re: TextSpanner terminates too early

[Trevor Bača]

On Wed, Nov 18, 2009 at 4:06 PM, Nick Payne <address@hidden> wrote:

Trevor Bača wrote:

On Wed, Nov 18, 2009 at 3:39 PM, Nick Payne <address@hidden <mailto:address@hidden>> wrote:

   Following example shows a problem I have in a score with multiple
   voices where I want the TextSpanner for a barre to be two bars
   long. Because the actual notes in the two bars are split between
   the voices, I have the \stopTextSpan on the silent eighth note at
   the end of bar two in voiceOne, but it is terminating as if I had
   it on the B flat at the start of the bar. In fact there is no
   difference in the output between having the \stopTextSpan on the B
   flat or s8. Why is this?

   %================================================================
   \version "2.13.7"

   barre = #(define-music-function (parser location fretnum dirn
   shorten adjBreak adjEnd)
     (string? number? pair? number? number?)
   #{
     \once \override TextSpanner #'bound-details #'left #'text =
     \markup { \small { \concat { "C" "." $fretnum " " } } }
     \once \override TextSpanner #'style = #'line
     \once \override TextSpanner #'font-shape = #'upright
     \once \override TextSpanner #'direction = #$dirn
     \once \override TextSpanner #'bound-details #'right #'text =
   \markup { \draw-line #(cons 0 (/ $dirn -1)) }
     \once \override TextSpanner #'bound-details #'left
   #'stencil-align-dir-y = #CENTER
     \once \override TextSpanner #'bound-details #'left #'padding =
   #(car $shorten)
     \once \override TextSpanner #'bound-details #'right #'padding =
   #(cdr $shorten)
     \once \override TextSpanner #'bound-details #'right-broken
   #'padding = #$adjEnd
     \once \override TextSpanner #'bound-details #'left-broken #'X =
   #$adjBreak
     \once \override TextSpanner #'bound-details #'left-broken #'text
   = ##f
     \once \override TextSpanner #'bound-details #'right-broken
   #'text = ##f
   #})

   \relative c' {
     \time 5/8
     \new Staff
     <<
     \new Voice = "first"
     { \voiceOne         \barre "6" #UP #'(-0.5 . -1) #5 #1
   bes8\startTextSpan f' aes d_( e) |
         bes'4 s4 s8\stopTextSpan |
     }
     \new Voice= "second"
     { \voiceTwo
         s2 s8 |
         bes f e aes, e' |
     }
     >>
   }
   %================================================================



Hi Nick,

I can't explain what's going on exactly, but transparent rests fix this one nicely:


%================================================================
\version "2.13.7"

barre = #(define-music-function (parser location fretnum dirn shorten adjBreak adjEnd)
 (string? number? pair? number? number?)
#{
 \once \override TextSpanner #'bound-details #'left #'text =
 \markup { \small { \concat { "C" "." $fretnum " " } } }
 \once \override TextSpanner #'style = #'line
 \once \override TextSpanner #'font-shape = #'upright
 \once \override TextSpanner #'direction = #$dirn
 \once \override TextSpanner #'bound-details #'right #'text = \markup { \draw-line #(cons 0 (/ $dirn -1)) }
 \once \override TextSpanner #'bound-details #'left #'stencil-align-dir-y = #CENTER
 \once \override TextSpanner #'bound-details #'left #'padding = #(car $shorten)
 \once \override TextSpanner #'bound-details #'right #'padding = #(cdr $shorten)
 \once \override TextSpanner #'bound-details #'right-broken #'padding = #$adjEnd
 \once \override TextSpanner #'bound-details #'left-broken #'X = #$adjBreak
 \once \override TextSpanner #'bound-details #'left-broken #'text = ##f
 \once \override TextSpanner #'bound-details #'right-broken #'text = ##f
#})

\relative c' {
 \time 5/8
 \new Staff
 <<
 \new Voice = "first"
 { \voiceOne         \barre "6" #UP #'(-0.5 . -1) #5 #1 bes8\startTextSpan f' aes d_( e) |
     bes'4
     \override Rest #'transparent = ##t
     r4 r8\stopTextSpan
     \revert Rest #'transparent
     |
 }
 \new Voice= "second"
 { \voiceTwo
     s2 s8 |
     bes f e aes, e' |
 }
 >>
}
%================================================================


Intuitively I always imagine skips as meaning "don't do any engraving work here." I don't know if that's actually an (or the) intended meaning skips or not, but it helps me remember that are a very few -- but important -- gotcha differences between skips and (transparent) rests, this situation with the spanners being one of them.

Wow, that was quick. Thanks, it does fix the problem. However, I went back and checked some other scores I have engraved, and in some of them I have used the same construction of TextSpanner sometimes starting and sometimes finishing on a spacing note, and there it does work correctly.

Is it possible that, in those other scores, there's some other engraved element (besides the skip) present at the beginning (or until the end) of the spanner that causes the context to be kept alive long enough for the spanner to engrave? In other words, could those scores be giving the appearance that the skip is enough to keep the spanner alive when, in fact, it's actual some other (visible) object that's doing the work of keeping the spanner alive? (But again, these are just intuitions here; I can't offer a solid explanation.)

Note also that if you start sprinkling transparent rests around to keep spanners alive that those transparent rests (like all transparent grobs) actually consume space on the page and can, therefore, push adjacent items around in dense configurations. Usually not a problem, but still something to keep in mind.

-- 
Trevor Bača
address@hidden