<direction> elements attached to grace notes?

[lemzwerg]

The specification says the following about the <direction> element.

For applications where a specific direction is indeed attached to a specific note, the element can be associated with the first element that follows it in score order that is not in a different voice.

A consequence of this sentence is that it should work for grace notes, too. However, neither Finale nor MuseScore seem to support that.

What is the 'official' point of view? For example, have a look at
the following MusicXML file.

If processed with LilyPond's musicxml2ly, I get the following result.

The import results of Finale (27.4.1.112) and MuseScore (a prerelease for 4.6.0) are as follows.

Is this a deficiency in the applications' import? Otherwise, in case I'm doing something wrong, what shall I do instead? How are wedges correctly attached to grace notes?

[mscuthbert]

Hi Werner -- I think your interpretation (along with that of musicxml2ly) is correct (maybe Michael or others also want to weigh in). When asking "but why does software [X] do [something else] instead?" Try to see what the limitations of that software are in general. For instance, in Finale 27 it appears that neither "smart shapes" nor expressions can be anchored to grace notes:

So if graces are not valid anchors internally then anchoring to the next valid note seems like a logical fallback.

In Dorico (at least 6.0) graces are valid anchor points for both letter dynamics (f, p) and wedges. Making a Dorico file like the following image produces a musicxml file that is very similar to the one you attached (24g) with the wedge and forte both before the first grace (though in different order).

but when imported back into Dorico, it has the same problems as Finale -- so it shows that Dorico believes this interpretation of MusicXML is indeed correct but just not fully supported.

(just to fully document, esp. if any Dorico devs are reading -- there were two further things on Dorico import that are worth noting: (1) the 24g file has no number="1" in the wedge direction start or stop - it is optional because there are no other wedges in the document - however Dorico will not import the wedge w/o a number. Adding it in, it imports. (2) as mentioned, the dynamics import as anchored the the main note, but they can be manually dragged to import to the last three gracenotes; however, they cannot be dragged to the first gracenote (at least on Mac) - there isn't an anchor spot. Alt-left can move it there with the keyboard. This is only a problem with gracenotes anchored to the first spot in the score).

[mdgood]

I agree with Myke that your understanding is correct and it's the software that's mistaken here. While you can anchor expressions (and I believe smart shapes) to grace notes in Finale, it involves a separate parameter that indicates how many grace notes are involved. This is tricky to compute during MusicXML import. Since this particularly notation is relatively uncommon, and export work was prioritized over import, it's not too surprising that this got missed.

[lemzwerg]

Thanks a lot for the confirmation, very much appreciated.

Your answers actually lead to a follow-up question that the MusicXML standard doesn't answer, AFAICS. What happens with <offset> if a <direction> element is attached to a grace note? This seems to be undefined – attaching to the next 'main' note instead doesn't make sense to me. Maybe a comment should be added that <offset> doesn't apply if <direction> is attached to a grace note, and only the default-x attribute can be used for such cases.

[mscuthbert]

I think if the standard went in that direction, then better than "<offset> doesn't apply" is that with directions attached to grace notes should always have <offset> of 0.

But I don't know that we want to be so prescriptive -- it would seem to be incorrect to put an offset of 3*divisions on a direction attached to a quarter note, but that has long been seen in real musicxml scores.

[mdgood]

Since grace notes and regular notes share the same "musical location" that offsets are referenced from, there is wiggle room for applications to interpret here. I think we should leave it as is, as this does not seem to have caused serious issues in practice. I think making this type of thing more precise seems better left to MNX at this point.

[lemzwerg]

OK, thanks again to both of you for your comments.