Bug 4984 correctly points out that xs:untyped should not be listed as a primitive simple type in Section 3.2. In addition, the types xs:dayTimeDuration and xs:yearMonthDuration should not be on the list either, as these are both derived rather than primitive simple types. Removing them would bring the number of primitive simple types to 21. I propose the following rewording: [Definition: There are 21 primitive simple types: the 19 defined in Section 3.2 Primitive datatypesXS2 of [Schema Part 2], xs:untypedAtomic and xs:anyAtomicType], defined in 2.6 Types.
It feels like a change like this can have dramatic effects, in particular on operator binding. Maybe it is as it is, for being somewhat aligned with the next version of Schema? xs:integer is neither a primitive type according to W3C XML Schema 3.2, but it's probably as well listed all over the specs as primitive. I have a vague memory of that either XQuery or F&O explicitly said that xs:integer was considered as a primitive type even though it isn't. In XQuery 1.0, H Glossary (Non-Normative), section xs:anyAtomicType are xs:integer and xs:untypedAtomic mentioned as primitive atomics.
These types aren't primitive, they are derived; but along with xs:integer they are specially recognized by the rules for operator binding and casting.
(In reply to comment #1) > > xs:integer is neither a primitive type according to W3C XML Schema 3.2, but > it's probably as well listed all over the specs as primitive. I have a vague > memory of that either XQuery or F&O explicitly said that xs:integer was > considered as a primitive type even though it isn't. You're maybe thinking of F+O section 17.1: "These four types [xs:untypedAtomic, xs:integer, xs:yearMonthDuration, and xs:dayTimeDuration] are not primitive types but they are treated as primitive types in this section." (Although I don't see why xs:untypedAtomic is "not primitive".) > In XQuery 1.0, H Glossary (Non-Normative), section xs:anyAtomicType are > xs:integer and xs:untypedAtomic mentioned as primitive atomics. Similarly in XQuery section 2.5.1, point 5. Getting back to the Data Model spec, section 2.6.2 (under xs:anyAtomicType) also mentions xs:integer as a primitive atomic type.
The WGs accepted the commentator's suggested replacement text at the 13 Nov 2007 telcon. Errata DM.E009.