Simon Albrecht <address@hidden> writes:
Hello,
consider the following example:
%%%%%%%%%%%%
\version "2.19.25"
\new Voice << { c''^~ c'' } { a'_~ a' } >>
\new Voice << { <c''^~> c'' } { <a'_~> a' } >>
{ <c''^~ a'_~> <c'' a'> }
%%%%%%%%%%%%
Contrary to (at least my) expectation the first example gives
tie-within-chord.ly:6:33 <0>: warning: Two simultaneous tie events,
junking this one
\new Voice << { c''^~ c'' } { a'
_~ a' } >>
and applies the first tie to both notes.
The other two give correct output, and it would serve consistency and
predictability if the first did also.
Possibly related: <http://sourceforge.net/p/testlilyissues/issues/2240/>.
No, even before the infamous issue 2240 ("Don't wrap EventChord around
rhythmic events by default.") an in-chord tie on an individual note and
an independent tie event would have behaved in that manner. Before
issue 2240, the input would have been translated into what is now
\new Voice << { <c''>^~ <c''> } { <a'>_~ <a'> } >>
\new Voice << { <c''^~> <c''> } { <a'_~> <a'> } >>
{ <c''^~ a'_~> <c'' a'> }
and the _result_ from typesetting your input is just like before issue 2240.