Is it helpful to have + at the end? I guess the question is whether e.g. if your last name is something Spanish, would you write <surname>Garcia</surname><surname>Lopez</surname> or <surname>Garcia Lopez</surname>? I guess it would be the latter, because that's not just less markup but also preempts unwanted name reordering.

Do you think this might work? (It's a bit of an issue that this codifies an order, but I don't know if that can be avoided.)

Is it helpful to have + at the end?

It's not from me, it's DocBook's content model for personname (minus the db._text pattern):

db.personname =
    ## The personal name of an individual
    element personname {
      db.personname.attlist,
     (db._text
       | (db.honorific
          | db.firstname
          | db.surname
          | db.lineage
          | db.othername)+
       | (db.honorific
          | db.givenname
          | db.surname
          | db.lineage
          | db.othername)+)
    }

The reason for that was to use DocBook's own content model.

I guess the question is whether e.g. if your last name is something Spanish, would you write <surname>Garcia</surname><surname>Lopez</surname> or <surname>Garcia Lopez</surname>?

Names are complicated. DocBook's markup is useful for maybe ~80-90% of all names. Not all names can be appropriately marked up. For example, some Gaelic names are even more difficult for DocBook.

Back to your question: I'm not familiar with Spanish names. It probably depends a bit on how the last name (or surname) is perceived. Are both parts independent from each other? Or are they an integral part and can't be separated? I don't know. 🤷‍♂️

Do you think this might work? (It's a bit of an issue that this codifies an order, but I don't know if that can be avoided.)

Technically, it does work. If you want to avoid an order, use the interleave pattern (&). The stylesheets picks the respective elements, see common/common.xsl, named template person.name.

If you add a @role attribute to personname, it can even pick and arrange the elements to format the name (valid values are family-given, last-first, and first-last). So the order of these elements are quite irrelevant.

I don't understand this -- "Geeko Gecko Chamaeleon" (given/given/sur) is a test case for a bad name, but "Geeko Gecko Chamaeleon jr." (given/given/sur/lineage) is a test case for a good name. Sound dodgy. :)

Good catch. Actually one of the givenname should be firstname.