NOTATIO

Excellent tisimst!
What I might think is like that (in Finale or I guess Sibelius):
1. count the greatest number of notes at the beat. At the beat 4 that is the Violin.
2. count extra "invisible" notes to fill the rest of the beat. In the case of Violin at the beat 4 it would be about 8 extra notes in the original provided.
3. Count all visible and invisible notes and add them as a tuplet, equally to all instruments.
4. Enter all visible&invisible notes, make smaller graces, separate real notes, remove flag and hide invisibles. And this is the "source":

OCTO wrote:Excellent tisimst!
What I might think is like that (in Finale or I guess Sibelius):
1. count the greatest number of notes at the beat. At the beat 4 that is the Violin.
2. count extra "invisible" notes to fill the rest of the beat. In the case of Violin at the beat 4 it would be about 8 extra notes in the original provided.
3. Count all visible and invisible notes and add them as a tuplet, equally to all instruments.
4. Enter all visible&invisible notes, make smaller graces, separate real notes, remove flag and hide invisibles.
shot 5.png

And this is the "source":
shot 6.png

That's very clever, OCTO!

I was thinking of something similar, but couldn't quite wrap my head around it.
Btw, you could also just use nested tuplets, which would eliminate the need to hide any notes.

Knut wrote: Btw, you could also just use nested tuplets, which would eliminate the need to hide any notes.

I have tried that, but it doesn't work. Or do you have another approach in mind?

OCTO wrote:

Knut wrote: Btw, you could also just use nested tuplets, which would eliminate the need to hide any notes.

I have tried that, but it doesn't work. Or do you have another approach in mind?

Like this:

Oh, I see, that is very smart!

tisimst wrote:Here's an excerpt of how LilyPond can handle this kind of situation. I wrote a little command to make it almost 100% automated (there are a few simple decisions to make, but it's straightforward) and safe against respacing.

Tisimst,

After testing OCTO's last approach, with my own modification, spacing still is a major issue, because using genuine is too high a value for the grace notes. I'm very curious what exactly you did to get such good results. Maybe the technique is possible to carry over to Finale in some way?

OCTO's solution (but not Knut's) hit me in the middle of last night with almost the same illustration, and here it is posted this morning! And I had a strong feeling that this would happen. We are often "of one mind" on this forum.

I was going to add the proviso that it is not absolutely certain to me that the grace notes are intended to line up exactly, and that the large notes must meet the small notes exactly as written, because it is not precise in the original. I think that Boulez may simply mean that each player should play the grace notes at about the same tempo and that the large note should come at the end, hitting in the order as shown in the score.

Knut wrote:I'm very curious what exactly you did to get such good results. Maybe the technique is possible to carry over to Finale in some way?

I'm not sure it is transferrable to other apps because I explicitly change the used duration of each note in the grace-main set without changing their appearance. My command takes several arguments that are manually entered:

A - Number of actual grace notes that will be used
B - Maximum number of grace notes in other staves
C - Duration the grace notes should fit within
D - Smallest grace note duration

A and D can definitely be more automated, and theoretically B, but it's a little beyond my coding capabilities at the moment. The B value allows simultaneous voices to line up their notes (if desired, like in the original post). The C value primarily controls how much of the main note's duration should be leeched off to the grace note group as a whole. The main note gets some of its duration back if A < B.

To be completely honest, it took me a while to get the math working correctly.

DatOrganistTho wrote:... in the source, did the "grace" notes remain "grace notes" in the code, or are they modified with magstep or scale?

I'm not aware of any way to make real grace notes begin on the beat, thus they are just normal notes that have been scaled down in appearance and with their durations adjusted per my above explanation.

John Ruggero wrote:I was going to add the proviso that it is not absolutely certain to me that the grace notes are intended to line up exactly, and that the large notes must meet the small notes exactly as written, because it is not precise in the original. I think that Boulez may simply mean that each player should play the grace notes at about the same tempo and that the large note should come at the end, hitting in the order as shown in the score.

I think you're right about that, but since grace notes simply should be spaced as close together as possible by default, lining them up (more or less) exactly makes sense. This also contributes to a more pleasing result, aesthetically.

tisimst wrote: The C value primarily controls how much of the main note's duration should be leeched off to the grace note group as a whole. The main note gets some of its duration back if A < B.

This seems to be the key to your success. As you point out, I can't think of a way to replicate this variable in Finale, unfortunately. There might be some alternate way to modify the note values via the tuplet tool that would give you the desired result, but I suspect that it would involve a lot of trial and error, as well as a headache.

Knut wrote:

I think you're right about that, but since grace notes simply should be spaced as close together as possible by default, lining them up (more or less) exactly makes sense. This also contributes to a more pleasing result, aesthetically.

At first, I was inclined to agree with you, but then thought more about it. Lining them up, if that is what we are trying to do, may give a misleading impression to the performers. I think that it is significant that the notes don't line up in the original. For that reason, I think that I would first use OCTO's or your system to get the notes approximately where they should be, and then hand adjust the result to match the original. Or I would try to come up with an even more complex set of tuplet relationships that would match what is on the page without hand adjustment. Or I would just wait for Dorico. Or maybe spring for Graphire Music Press and see what that could do.