As part of the evolving family of XML standards, XPath 2.0 MUST support the W3C XML architecture by integrating well with other standards in the family.

XPath 2.0 data model MUST be expressed in terms of the XML Infoset, including Post Schema Validation Infoset contributions. Ideally, XSLT, XPath, and XML Query should share a common data model.

XSLT 2.0 contains an expression language which is "XPath 2.0 plus XSLT 2.0 extensions". XML Query 1.0 contains an expression language which we believe should be "XPath 2.0 plus XML Query 1.0 extensions." The scope of XPath 2.0 must be the set of common functionality between the expression language of XSLT 2.0 and the expression language of XML Query 1.0. This will guarantee for Web developers and users that any common functionality are implemented with identical syntax and semantics.

It MUST be possible for boolean expressions involving node-sets to explicitly use "for any" or "for all" semantics, otherwise known as explicit existential quantification and explicit universal quantification, respectively. Any implicit quantification should be explained in terms of these explicit semantics.

XPath 1.0 currently supports sum() and count(). XSLT users have frequently requested the addition of min() and max(). XPath 2.0 MUST address this by extending its basic set of aggegration functions by adopting some or all of the ones defined for XML Query.

Any valid XPath 1.0 expression SHOULD also be a valid XPath 2.0 expression, and have the same semantics when operating in the absence of XML Schema type information. If not possible to achieve due to other requirements, XPath 2.0 should minimize the number of changes to the XPath 2.0 syntax to maximize backward compatibility with XPath 1.0.

XPath 1.0 supports the union of two node sets. Set functionality in XPath 2.0 SHOULD be expanded to include intersection and difference functions, and should be compatible with equivalent functions supported by XML Query.

XML Schema allows decimals to have a leading plus. To align better with this, XPath 2.0 SHOULD allow this by providing a unary plus operator.

Users of XPath 1.0 have requested enhancements to simplify expression of common XPath use cases. XPath 2.0 MUST address these frequently requested enhancements.

To better align with XPointer and to simplify the use of XPath expressions in which multiple alternatives are allowed for a given location step, XPath 2.0 MUST relax current restrictions for what can appear after a '/' in an XPath expression. Neither unions nor node-set functions are allowed to appear after a '/' in XPath 1.0.

Many users have requested the ability to return a conditional value based on a boolean expression. XPath 2.0 MUST provide a conditional expression which takes three expressions:

and evaluates to expression2 if expression1 is true and to expression3 if expression1 is false. Only expression1 and the expression to be returned must be evaluated.

In XPath 1.0, the use of a collection-valued subexpression can introduce an implicit existential quantification or choose-first-member operation into the containing expression's semantics. XPath 2.0 MUST define a consistent implicit semantics for expressions that have collection-valued subexpressions. This may require that XPath 2.0 expressions use explicit quantification or indexing expressions to achieve the same implicit semantics provided in XPath 1.0.

Users of XSLT 1.0 have requested additional string manipulation functions. Common requests include string padding (with spaces, dashes, or other characters), string replacement, and converting strings to upper and lower case. XPath 2.0 SHOULD provide additional string functions. As with any string functions, internationalization issues need to be addressed to insure that the functionality is as broadly useful as possible.

XPath 1.0 provides the translate() function that allows each single character in a source string to be translated to another single character in the result string, or to be removed. Users of XPath 1.0 frequently want to replace sequences of consecutive characters with other sequences of characters, which translate() does not support. XPath 2.0 SHOULD support a more flexible string replacement function.

Often string values need to be padded on the left or right to make the value occupy a fixed length. XPath 2.0 SHOULD support a string padding function that permits any character as a padding character.

XPath 2.0 SHOULD provide the ability to convert the case of text to upper or lower case for presentation and/or comparison.

Users of XPath 1.0 frequently request the ability to apply an aggregate function, like sum(), to the values of expressions applied to a node set. XPath 2.0 SHOULD support aggregation functions over expressions applied to node sets.

Regular expressions provide a powerful way to specify string pattern matching and now play an important role in XML Schema as the mechanism by which pattern facets are specified. XPath 2.0 MUST support regular expressions for matching against strings using the regular expression notation established in XML Schema: Datatypes.

XML Schema: Datatypes defines a set of primitive datatypes. In addition to the types supported by the XPath 1.0 data model, string, number, boolean, and node-set, the XPath 2.0 data model MUST support XML Schema primitive types. XPath 2.0 will extend the XPath data model to accommodate working with elements and attribute values having an XML Schema primitive type.

XPath 2.0 MUST support the operators and type-coercion rules defined by the joint XSLT/Schema/Query task force on operators.

XML Schema specifies a lexical representation for doubles and floats that includes scientific notation as well as INF, -INF or NaN. XPath 2.0 MUST support the lexical representations of floats and doubles supported by XML Schema.

XPath 2.0 MUST define an appropriate set of functions to allow users to cast and construct instances of XML Schema primitive types. At a minimum, this set MUST include cast and constructor functions for URI and date/time types.

XML Schema allows the definition of simple types derived by list, including lists of unions of non-list simple types. XPath 2.0 SHOULD support an ordered list of simple-typed values.

The XPath 1.0 type system supports the number, string, boolean, and node-set types. XML Schema: Datatypes introduces many new types. XPath 1.0 supports conversion of the simple-typed values of elements and attributes to strings. In addition to this functionality, XPath 2.0 MUST support access to the native, simple-typed value of an element or attribute.

Since the typed value of an element can be null, XPath 2.0 MUST define how the behavior of operations applies to null values.

XML Schema: Structures enables users to define structured types and associate them to elements in a schema. XPath 2.0 SHOULD provide support for the common operations needed for navigation and selection of typed elements.

XML Schema : Structures provides the ability to define the type of an element or attribute. XPath 2.0 SHOULD be able to test whether an element or attribute is an instance of a given type.

XML Schema : Structures provides the ability to define a hierarchy of types by derivation. XPath 2.0 SHOULD be able to selects elements or attributes that are instances of a type, also matching any types derived from it by restriction or extension.

XML Schema: Datatypes provides the ability to include two or more element names in a substitution group. XPath 2.0 SHOULD be able to test whether an element is a member of an XML Schema substitution group.

XML Schema: Datatypes supports named, multi-part keys. XPath 2.0 SHOULD support a mechanism for looking up the element to which a Schema key refers. Similar mechanisms already exist, such as the id() function in XPath 1.0 and the key() function in XSLT.