> Actually, Geoff was not proposing (although I don't want to put words in
> his mouth) mixing tags and PCDATA.  What I believe he was proposing was
> either or.  According to a DTD schema both are considered mixed content
> models.  Here is his <creator> example:
>   <creator>
>     <type>personal</type>
>     <name>Abrams, Michael</name>
>     <description>(American artist, 20th c.)</description>
>   </creator>
> In DTD term he was proposing:
>   <!ELEMENT creator (#PCDATA|(type,name,description)>
> so you can have either any text you want or the elements type, name
> and description.  In this model you cannot mix text between the
> elements type, name and description.  He was not proposing:
>   <!ELEMENT creator (#PCDATA|type|name|description)>
> The latter would be like HTML where you can do:
>   <div>
>     This is some text...
>     <a href="#">This is a link</a>
>     This is more text...
>   </div>

Sorry if there was any confusion about my proposal. Priscilla is correct in
her interpretation. Elements that may contain sub-elements may not contain
PCDATA (ever) and vice versa. The "creator" field would be defined partially
like this:

  <!ELEMENT creator (type,name,description)>

Whereas the "name" field would be defined thus:


It would be much easier for someone searching and parsing a MODS document to
not have to check both the "creator" element and the "name" sub-element for
PCDATA.  I want the data to be as structured as possible and at the same
time, reduce the number of decisions a cataloger must make when creating a
record.  It's better for both the producer and consumer of this data.