Lily and petrucci (whaddapair!)

[William R. Brohinksy]

Heyla! 

OK, always being one to tilt at windmills, I'm continuing with my
transcription of Harmonice Musices Odhecaton A. Lilypond is doing great
for this: I've gotten to the point where I've got a routine: mark off
the original notation (on a xerox, of course) at each Longa, and
transcribe that to four measures per line. I've been putting in ~'s as
needed to get notes that cross breves (each measure is a breve,
transcribed to 2/2 from slashed-C, transcriptions from O are always more
complicated to deal with)and using the fantastic array of modern clefs
to do the original setting and proof-reading without having to cross
from, say, mezzo-soprano clef to treble clef or alto clef as I try to
proof.

All to the positive, but of course, there's always another side to the
story. Eventually, I will want to try a barless version, at which point
having all those ~'s becomes a liability: they'd have to be translated
back to dotted notes or note values that cross 'measures'. Now, I've
also discovered, by perusing ancient-fonts.ly (in
/usr/lilypond/input/test/) that it's not all that hard to make mensural
noteheads and center stems. This gives rise to a whole wishlist, by the
way, as well as a bug report:

Wishlist for edited mensural notation:
{of necessity, in two parts}
For duple modus/tempus/prolatio/:
A mode in Lilypond that allows notation without barlines and without
complaining of dying if a note value exceeds the current accrued measure
size is going to be necessary. In short, note values should be used to
align notes vertically in score formats, and totted up to ensure that
ends-of-line are achieved based on longa, breve or semibreve divisions.
(I've noticed that the renaissance scribes and publishers didn't really
take care to divide lines in partbooks based on anything to do with the
music, splitting here on long boundary, there on breve boundaries, and
somewhere else in the middle of a breve or even the middle of a
semibreve. Since I'm interested in well-edited versions, rather than
merely trying to make Lily output a typeset facsimile of some original
scribe's note-stacking.) The choice should be an option. 

An alternative to this would be a notation that allows music to be
parsed for barlines (ie, the ~ becomes a suggestion of where a compound
note can be split) but when parsing for barless, adds up the values
linked by ~'s and sets one note at that pitch, sized to include the
entire duration. This doesn't have to be too bulletproof: if you enter 5
semibreves, all ~'d together, shame on you (since there isn't a
reasonable way to write that note value without a tie, and renaissance
music didn't have ties!)

Perfect handling (modus perfectum/tempus perfectum/prolatio perfectum)
White mensural notation's perfect systems are complicated and only
sometimes determinate. Rather than wanting a system that you feed the
values on the paper into, tell what level (mode, time or prolation) is
perfect, and let the machine crank out the translation, I'd prefer a
'ghost note' mode: perhaps *a2 notation or some such. To illustrate the
problem, In perfect tempus, breve semibreve semibreve breve breve as a
unit becomes 
3/2 dottedwhole | half whole | dotted-whole | dotted-whole |
and if you had wanted to notate 
3/2 whole half | half whole| dotted-whole |
you would have to put a "punctus divisionis" between the two semibreves:
breve semibreve . semibreve breve breve
Just about everyone who has been through music school has been told
about this at some level of glossing: in the renaissance, binary systems
always had binary values, but trinary systems sometimes assigned to the
same note shape binary and other times trinary values. 

Instead of complicating life by trying to figure out all the rules and
build them into Lilypond, it's better IMHO ( being one of the people who
has spent a lot of time figuring out how to do it for himself) that we
be able to notate that first phrase as
a1 *a2 | g2 f1 | e1 *e2 |  d1 *d2 |
and have the system understand that for barred systems, this is read as
meaning 
a1. | g2 f1 | e1. | d1.|
and in barless modern notation, as 
a1. g2 f1 e1. d1.
and in barless mensural notation as
a1 g2 f2 e1 d1
Which one is used should be independently controllable, so that if I
have to write the simplified version for someone who is just beginning
to read original notation (in a tutorial, for instance) I can still get
the barless modern notation version on ancient fonts!



A larger ancient font notehead, or a new spacing regime. The current
lozenge-heads just seem too small and insubstantial unless you use
paper26, and frankly, having a 2-page 16-point  modern notation piece
grow to 4 pages at 26 really isn't good. Also, they are _quite_ square!
Both Petrucci (whose Odhecaton and Canti B and C show us a font that is
balanced and beautiful while still being emminently readable) and
Dijon517 show noteheads that are taller than they are wide, except for
the breve and long which are pretty square (and not canted like the
semibreve!) Also, there's the Maxima, which resembles what is present in
Lilypond's fonts as the \long now. (the breve is square or perhaps just
a bit less wide than tall, the long is a breve with a tail, and the
maxima is a long that is 2 to 3 times wider). (As for paper26 4-page
scores, I've rarely seen renaissance musicians willingly sharing a
stand, even with large-font scores. Possibly there's a psychology paper
here for someone!)

An ancient-font semiminima flag (eighth not-like flag). Using the modern
notation flag makes poor balance.

A coloration regime which allows for the 'flipping' of note color, breve
for blackened breve, semibreve for blackened semibreve, minim and
semiminim, fusa and whited fusa, etc. Likewise, I'd love to have a way
to notate a plain bracket over notes that had originally been ligitures.
"Just what we needed", you say, "another couple of grouping
punctuation". Yeah, I say. Just exactly. (Frankly, since there are no
slurs in White Mensural Notation, we could use the slur brackets for
ligatures!)

More clefs, or a clearer way to set up the ones we have. While the C
clefs are available on many lines, the F and G clefs aren't, and should
be. Specifically the ones I want are mensural1_F on any of the top four
lines, mensural2_g on the bottom three lines, and mensural3_c on all
lines While even I would have trouble justifying mensural_g on the top
line, G on four lines, and F and C on five are a logical approach. And
if the flags make mf-work desirable, I would love to have the clefs made
to look a bit more like petrucci's (maybe time to add petrucci1_c/f/g?).
They're lovely, consistant, and very personable. If Lilypond wants to
typeset beautiful music, then for mensural notation, Petrucci is where
to get your inspirations!

Bug report on ancient-font.ly output in Lilypond 1.4.4:
The compiler complains every time it encounters 
\property Voice.Stem \override #'stem-centered = ##t
saying that it has no idea what stem-centered is. However, it does print
centered stems!
Additionally, it seems that 
\property Staff.StaffSymbol \override #'line-count = #4
is only half-obeyed. There are no objections to parsing it, and
vaticana_fa1 falls on the third space down...but there are five lines.
Apparently the fifth (first if you count from the bottom) isn't noticed
by whatever places the clef, but it's printed.

In summary, what I'm looking for are:
-new ways of keeping track of note values which are based less on
measures than on finding
a good place to split barless scores, but which still align notes
vertically in a method that makes reading and understanding easier.
-a better ancient font which benefits from the balance and placement
rules obvious in the better of original sources
-a better set of ancient font clefs, at the very least allowing the
mensural clefs to be placed on any line. (placement of the Cclef on the
third space up would be interesting... but not something I'm asking for)
and the possibility of a new clef set which is based on Petrucci's
fonts.
-a mode that is useful in both ancient and modern font settings, which
allows either {entering music without consideration of barlines and
splits compound (bar-crossing) values automagically when needed} or
{entering music with ~'s which are taken like barchecks now: suggestions
in barred layouts, and count-ligitures in barless layouts} resulting in
a single input producing a proper output in barless/non-tie and
barless/tie-over-bar modes.
-ligiture braces and coloration grouping indicators, which place
'rectangular braces' (like truplet brackets without numbers, and
indicated by start and stop, rather than tuplet values) over notes that
had been in ligitures and 'flip' displayed forms of notes where
coloration is wanted. (Maybe a \color variable to set when using color
printer output? 8^)

and of course,
- things like the number of lines should actually be displayed properly!

I suppose this should be queued up for 1.6 or 1.7?

raybro