You're close, if you omit the asterisks. Examples below:Normally the entire not given name part should be in GEDCOM in a way that others import whole. If not, there is a bug, and please post a bug ticket (if bug is not on Rootsmagic side).
Avoiding features due to bugs does not allow the bugs to be fixed.
If I remember correct, in Gedcom you can store John /*FAMILY NAME PART*/ and it is then up to the software to include it.
1 NAME Ilse /de Lange/ 2 GIVN Ilse 2 SPFX de 2 SURN Lange1 NAME Fatima /Moreiro de Melo/ 2 GIVN Fatima 2 SPFX , de 2 SURN Moreiro, MeloFor Ilse, any software that reads the name parts can rebuild the full name by concatenating the parts, but for Fatima that doesn't work. GEDCOM 5.5 does not allow for multiple name pieces, so IMO the only proper way to export Fatima's name is to reassemble all surname parts before export. I will put that in my bug report.
For my great great ... grandfather, results are like this:
1 NAME Andries Dircks /Noy/ 2 GIVN Andries Dircks 2 SURN Noy1 NAME Andries /Dircks Noy/ 2 GIVN Andries 2 SPFX , 2 SURN Dircks, NoyWhere the 1st is how most Dutch genealogists enter names, and the 2nd what you get with multiple names.I agree, and name parts are already part of the gedcomx draft. In fact, it looks like the types that Gramps supports are either taken from that gedcomx draft, or the other way around.
As you mention, you can use Gramps in this way also by not clicking the multiple surname button.
Name support is one reason Gedcom should be improved. For older names, there is a lot of information in the names, and you want to store that information in a way that can easily be retrieved for analysis. Just adding a not to document the information on the name is a weak way to do this. Storing a name is patronimic, or how a Portuguese name splits can be good for your research later on.
But for now, most of us have to live with GEDCOM 5.5, and realize that name parts won't survive, unless we take appropriate measures.