Incorrect rendering for multi-voice Clef Change

[arshiacont]

Unit test:

{[ \staff<1> 
   (* meas. 1 *)  \clef<"g2"> {e/1, b/1  }
 \bar
   (* meas. 2 *)  \clef<"f4"> \restFormat<dy=-2>( _/4)
 \clef<"g2">  {e/2, g/2, c2/2  } {e1/4, g/4, c2/4  } ]
 , 
[ \staff<1> 
   (* meas. 1 *)  \clef<"g2"> empty/1 \bar
   (* meas. 2 *)  \clef<"f4"> {e-2/1, e-1/1  } ]
  }

leads to:

whereas the second voice alone is

corresponding to:

{[ \staff<1> 
   (* meas. 1 *)  \clef<"g2"> empty/1 \bar
   (* meas. 2 *)  \clef<"f4"> {e-2/1, e-1/1  } ]}

[arshiacont]

I thought for a second that the behaviour is coherent sincethe F clef part of Voice 2 spans the "G CLEF" part of Voice 1. But reducing that part to the clef span results to the same behaviour:

{[ \staff<1> 
   (* meas. 1 *)  \clef<"g2"> {e/1, b/1  }
 \bar
   (* meas. 2 *)  \clef<"f4"> \restFormat<dy=-2>( _/4)
 \clef<"g2">  {e/2, g/2, c2/2  } {e1/4, g/4, c2/4  } ]
 , 
[ \staff<1> 
   (* meas. 1 *)  \clef<"g2"> empty/1 \bar
   (* meas. 2 *)  \clef<"f4"> {e-2/4, e-1/4  } empty/2 empty/4 ]
  }

[dfober]

I strongly discourage the use of keys in this context. In fact, it would seem that it is necessary to put at least one note in F clef on the staff 1 to correct the problem. But the notation can become very ambiguous:

{
[
	\clef<"g2"> { e/1, g }
	% the following uses the f clef (that is on the second voice)
	_/4 e-0/4
	% and back to explicit g clef
	\clef<"g2">  { e1/2, g, c2 } { e1/4, g, c2 } 
] , 
[
	\staff<1>  empty/1 \bar
	\clef<"f4"> { e-2/4, e-1 }
]
}

[arshiacont]

Then what would be the correct way to write such a score without adding a new note that the composer hasn't written?? :)
To me, there should be some sort of Clef/Position check in the AR to determine the correct clef at each time-position between voices to avoid this.
In fact, this is how I manage to do it in xml2guido (I do book-keeping of (clef, position) for each voice in a staff).

[dfober]

it would seem that it is necessary to put at least one note in F clef on the staff 1

was not intended as a solution, just as a debugging remark. But the ambiguity in the notation remains.
Sorry for the confusion.

[arshiacont]

No problem. Let me know if I can contribute / test.

[arshiacont]

Notes: Behaviour only on chords.

[arshiacont]

Another example for the same problem where the "chord" annotated with (* PROBLEM*) in measure 2 is being rendered in a different clef than indicated:

{
[ \staff<1> 
   (* meas. 1 *)  
   \clef<"f4"> \key<2>  
   \stemsUp d0/8. 
   \clef<"g2">  \tieBegin:1<curve="down"> d1/16 d/8 \tieEnd:1
  \tieBegin:1<curve="up"> b/2. \bar
 
   (* meas. 2 *)  \stemsUp \beamsOff b/4. \tieEnd:1 
   \tieBegin:1<curve="up"> b/4. b/8 \tieEnd:1 
   \clef<"f4"> 
   empty/8 empty/8]
 , 
[ \staff<1> 
(* meas. 1 *)  empty*3/8 \stemsDown 
\beamBegin:1 d/8. 
 d/16 b0/8  \beamEnd:1 \tieBegin:1<curve="down"> d1/4. \bar
(* meas. 2 *) d/4. \tieEnd:1 
\stemsDown \beamBegin:1 b0/8. d/16 b0/8 \beamEnd:1  
 { d1/8., f#/8.  } (* PROBLEM *)
  \beamBegin:1  f#/16 f#0/8\beamEnd:1 
  ]
}

NOTE: This only happens if that event is a Chord... . If I change it to a note, everything is gonna be fine (here changing that chord { d1/8., f#/8. } to d1/8.:

@dfober If you guide on where to look, I might be able to fix this as I'm getting more and more acquainted with the library! ;)