I've searched both the list archives and LilyIssues and do not see this reported.
I use temporary polyphonic contexts a lot in my choral compositions. I see that using any version of 2.19 breaks lyric placement if there are slurred notes in /voiceOne against which lyrics should be placed. There
are some regression testing examples for polyphonic contexts and
lyrics, but they do no use the following lilypond structure which I have
been using for several years:
<<
{ \voiceOne }
{ \new Voice { \voiceTwo } }
>> /oneVoice
In this example in version 2.18.*, for the word "waters," the syllable "wat" is engraved on the measure 2 slurred "gs'2( a')" with "ers" on the a'2 of the next measure. In 2.19.* the slur is not factored in the lyric placement and "ers" is engraved on the a' in measure 2 and not in measure 3.
<code>
\language "english"
test = {
<a' fs'>2 <cs'' fs'>4 <cs'' fs'> |
<< { \voiceOne
gs'2( a') |
a'2. s4
}
{ \new Voice
{ \voiceTwo
e'1 |
e'2. b'4\rest
}
}
>> \oneVoice \bar "||"
}
verseTestVoice = \lyricmode {
side the still wat -- ers.
}
testVoicePart = \new Staff { \global \test }
\addlyrics { \verseTestVoice }
\score {
<<
\testVoicePart
>>
\layout {
}
}
</code?