Re: Stem lenghts

[Mark Polesky]

Mark Polesky wrote:
> Pekka Siponen wrote:

> > Also the stem lengths on the third and fourth 
> > beat should be the same. The third beat stems
> > are a little too long and the fourth beat stem
> > is too short. Attached is the lilypond output.
> 
> I will look into this. What source are you using
> to make these assertions?

Your assertion that the 4th beat stem is too short
is supported by Ted Ross, who says on p.85 that the
downward stem on a middle-line note is "normally"
3.5 spaces (one octave) tall but also that "these
stems are quite often more effective at 3.25 
spaces". The stem in your example is clearly only 3
spaces tall, so perhaps a bug report is appropriate.

___________________________________________________


The situation in beat 3 is *much* more complicated,
but again, you can cite Ted Ross in your defense.
On p.121, he writes:

When the stems are pointing down:
  a. The primary beam will either sit on or
      straddle a staff line. If the beam is
      slanted and one end sits, the other must
      straddle a staff line and vice versa.
  b. The secondary beam will either straddle or 
      hang from a staff line. If the beam is
      slanted and one end straddles, the other
      end must hang on a staff line.
  c. If the primary beam straddles, the secondary
      beam must hang on a staff line. If the
      primary beam sits, the secondary beam must
      straddle a staff line.
  d. A primary beam cannot hang from a staff line
      - a secondary beam cannot sit on a staff
      line.

However, I can't find any information in his book
to answer the question: when he refers to the end
of a secondary beam, does he mean: where the beam
would end if it encompassed the same span as the
primary beam? Because in your case the right edge
of the secondary beam is to the left of the right
edge of the primary beam.

In either case, it does look like LilyPond's 
default violates guideline b because the right
edge of the secondary beam neither straddles nor
hangs, nor would it if it encompassed the same
span as the primary beam (ie. if the notes D-C-B
were all 16th notes). In that case, the right edge
would sit on the (nonexistent) first ledger line
if the beam were extended to the B stem. I should
point out that Ross considers these nonexistent
ledger when determining beam slopes, as can be
seen in his examples starting on p.104.

As far as I can see, the only possible solution
that follows all of these (admittedly complex)
rules is to make the D stem 4 spaces tall and the
B stem 3.25 spaces tall. Then the left edge of the
primary beam would sit on the invisible first
ledger line and the right edge would straddle the
same ledger line. The left edge of the secondary
beam would straddle the first staff line and the
right edge would hang from the same line (if it
were extended all the way to the B stem). I've 
labeled this solution "A" in the attached code.

If, however, we take Ross's "beam end" to mean its
actual end, then the rightmost stem would need to
be a quarter-space lower (labeled "B" below").

As far as implementing this, I have no idea. I
don't know how LilyPond is set up to calculate
these things, and I imagine making changes of this
sort would be quite involved. I'll leave that to
the developers.

- Mark
___________________________________________________


\version "2.13.0"

{ 
  d''16^"default" c'' b'8
  \once \override Beam #'positions = #'(-2.75 . -3)
  d''16^"A" c'' b'8
  \once \override Beam #'positions = #'(-2.75 . -3.25)
  d''16^"B" c'' b'8
}

stem-lengths-comparison.png
Description: PNG image