Re: lengthening broken ties

[David Nalesnik]

Hi Werner,

On Sun, Feb 15, 2015 at 8:09 AM, Werner LEMBERG <address@hidden> wrote:

>> If it's convenient, have a look at the attached patch/sketch.  It
>> adds a property, minimum-length-left-broken, which lets you adjust
>> broken bits that start a line.
>
> Excellent!  What an improvement with just a few lines of code!
> Thanks a lot.

Now that the change is in the repository, I've recompiled lilypond to
try it out with real-world code.  Unfortunately, it doesn't work as
expected.  Consider this snippet (together with the attached image).

  \paper {
    line-width = 50\mm
  }


  \relative c'' {
    <bes des e g>1 ~ \break q \break

    \once\override Tie.minimum-length-after-break = 10
    <bes des e g>1 ~ \break q \break

    \time 4/4 <bes des e g>1 ~ \break \time 3/4 q2. \break

    \once\override Tie.minimum-length-after-break = 10
    \time 4/4 <bes des e g>1 ~ \break \time 3/4 q2. \break
  }

In bar 2 you can see the effect of `minimum-length-after-break'.

You mean bar 4--bar 2 is the default.

 

However, in bar 8, nothing changes!  Value 10 is obviously too small
to get an effect if the line starts with a time signature – it seems
that the horizontal space of the time signature must be taken into
account also.

Any chance to fix this?  BTW, if you compare bar 2 with bar 6, you see
exactly the opposite effect w.r.t. the default tie length: with a time
signature the tie looks OK but without it is far too short...

 If you change 'minimum-length-after-break to 'minimum-length, you'll get exactly the same result with the second half of each tied chord.  This isn't a problem with the patch, as I changed nothing relating to the reference points from which the distances are reckoned.

From your analysis it appear that LilyPond is taking the entire NonMusicalPaperColumn into account rather than just the right extent.  I'll have a look.  (This could be major!)

--David