Re: wrong position of a tie between notes in chord

[Janek Warchoł]

On Wed, Feb 29, 2012 at 4:38 PM, Pavel Roskin <address@hidden> wrote:
> On Wed, 29 Feb 2012 14:47:48 +0000 (UTC)
> Karol Majewski <address@hidden> wrote:
>
>> % Tie that should connect 'g' and 'g'
>> % is located very low, so that it looks like a tie between 'd' and
>> 'es'
>>
>> \version "2.15.30"
>>
>> #(set-default-paper-size "a4")
>> #(set-global-staff-size 18)
>>
>> LiniaJeden =  \relative c'' {
>>   \clef "treble" \key c \major \numericTimeSignature \time 3/4
>>   << { g'4 f2 } \\ { < d, g ~ c ~ >2. } >>
>>   << { g'4 c2 } \\ { < es,, g c >2. } >>
>>
>> }
>>
>> \score {
>>     \new Staff = "LiniaJeden" \LiniaJeden
>> }
>
> Lilypond 2.14.2 did a better job, even though its output could be
> improved tie.
>
> I understand that Lilypond shifts the tie lower to avoid collision
> with the accidental, and it would look OK if there were no notes below
> that are not tied.
>
> It's tempting to call it a regression.  However, I don't think Lilypond
> has logic to make ties unambiguous, so the fix would be either to
> revert collision avoidance and decrease quality of other scores, or to
> add such logic, which might be a difficult task.

yup... since half a year i'm writing a detailed and systematic report
on LilyPond ties, which includes cases like this.  However, i'm so
much absorbed by the other stuff (like preparing Google Summer of Code
application for LilyPond) that i cannot finish it.
:(

thanks,
Janek