Is this editorial?

I don't see this as problematic: it's conventional with value-semantic object types to refer to objects as having properties when we actually mean that the value of the object has that property. Given:

int i = 42;

I think the statement "i is even" clearly conveys the meaning "i has the value 42 which is even."

I thought that defining

• past-the-end iterators
• dereferenceable iterators
• singular iterators

and putting them in the index may improve the readability of the Standard.

All three terms are defined for iterator values in [iterator.requirements.general]/7, although not properly indexed. Would it suffice to index those terms?

Would it suffice to index those terms?

Yes, that works.

• While the term "invalid pointer value" has the two index entries,

  • invalid pointer value
  • value -> invalid pointer

  I think it is more useful if the term "xxxx iterator value" has the following index entries

  • xxxx iterator value
  • iterator -> xxxx iterator value (instead of value -> …)

• In the Standard (unlike the SGI doc), singular iterator values are only used in "22.3 Iterator Requirements". Thus, since the term "singular iterator value" meets this,

  As a rule of thumb, if a term is local to a subclause, do not make it italics and do not index it

  it seems better not to index it.

The linked pull request will consistently use "xxxx iterator" in the index, not "xxxx iterator value", following the argument of @CaseyCarter. However, the actual definitions are not changed.