The quirkiness of the definition derives not from the definition

itself, because there may well be justification for the
existence of a node type that encompasses text with embedded
comments. Thats OK.

text nodes are leaves: they never have embedded comments. Your complaint
appears to be just a complete misunderstanding of what text() returns
(which is never anything with a comment in it). It's OK to misunderstand
the language but rather odd to assert that your misunderstanding is due
to a failure in the language definition.

Perhaps you'd want it to do what string() does but that's called
string() not text():-)

You see there right where the smiley is, is where we differ. I don't
see that as anything to smile about. To me thats just thats a smart
alec distinction.

The distinction is fundamental to the use of XPath. text() returns a
sequence of text nodes, you can ask for their identity, their parent and
sibling nodes etc. string() (which appears to be the function you
wanted) returns a _string_ which is an atomic value dissociated from any
document tree. The distinction between atomic values and nodes pervades
all uses of XSLT and XPath.