Re: Constructive Criticism and a Question

[Trevor Bača]

On 1/2/07, address@hidden <address@hidden> wrote:
> > ... in irregular, tuplet-intensive music it may be sensible to create a
> > music function for sequences of tuplets. In addition, it's IMHO a more
> > lilypondesque solution than tupletSpannerDuration, once we support fractions
> > as music function arguments.
>
> If I understand you correctly, this would involve specifying, one way or
> another, the duration of each actual tuplet.  Explicit specification of a
> duration (other than by an external tupletSpannerDuration declaration) has
> been suggested by another user, and IMO it would be a good idea, although
> I gather that Han-Wen is not in favour of the idea.
>
> But I have a question about how one would specify a duration.  Specifying
> durations in the way we usually think about them allows actual durations
> that look like this:
> 1    ==> 1
> 2... ==> 15/16
> 2..  ==> 7/8
> 2.   ==> 3/4
> 4... ==> 15/32
> 4..  ==> 7/16
> 4.   ==> 3/8
> 4    ==> 1/4
> (etc.)
> so that only durations of the form
>    2^(p-1) / 2^q  (where p < q)
> can be specified this way.  But given the extravagancies of contemporary
> music, wouldn't it be possible, for example, to have a tuplet where 4
> eighth notes would be played over a time interval of 5 eighths --
>       \times 5/4  {c8 d e f}
> Or does such a thing never happen?

It most certainly does. All the time, in fact.

Not only that, but an increasing amount of modern music considers the
following valid:

 \times 5/7 { c'16 c'16 c'16 c'16 c'16 c'16 } % note six sixteenths
only (not seven)

What's going on here are 7 sixteenths in the time of 5 sixteenths BUT
only the first 6 of those 7 sixteenths actually appear. You could call
this sort of thing a "broken" tuplet, and the overall duration of the
figure here is then 6/7 of 5 sixteenths or 15/56 of a whole note
(which lily would express as #(ly:make-moment 15 56)).

These broken tuplets would then pose a much, much greater difficulty
of expressing were we to move to a duration-based syntax for tuplets.

Broken tuplets may look insane, but take a look at the devestatingly
beatiful Sciarrino flute pieces -- plenty of examples, and all
completely idiomatic.

Note, importantly, that, with the present tuplet syntax, lily handles
all tuplets -- *including broken ones* -- correctly out of the box.
This sort of thing brings Finale and Sibelius screaming to their
knees. (This seems to be an extension of the fact that lily gets one
thing *exceedingly* correct: the duration model of musical time. Out
of the box you can also specify time signatures like 6/15, 5/28, 3/10
and so on, all of which bring other musical notation programs -- with
the the notable exception of SCORE -- to a crashing standstill. Or at
least the last time I bothered to check.)

I've been watching the tuplet discussion with some hesitation. I think
chaning \times to \tuplet is a great idea for the reason that started
the thread: \times is too close to \time. But it seems to me that most
of the suggestions following that initial suggestion begin to confuse
the essential time-scaling function of tuplet brackets (which is their
absolutely core purpose, both in the common practice and now) and
other graphical aspects of the notation such as beaming, grouping (and
even accentuation). Beaming and grouping are terribly important, of
course, but I think that it's best to leave their specification out of
the core tuplet syntax.

More important is to fix the fact that

 \times { c8 d e f }

will currently by default print with only a 4 in the tuplet bracket,
which is mathematically wrong; the denominator 5 must appear.


--
Trevor Bača
address@hidden