27455 2014-11-27 11:16:46 +0000 [xp31[ Unary lookup: specification gaps 2014-12-15 01:13:37 +0000 1 1 1 Unclassified XPath / XQuery / XSLT XPath 3.1 Last Call drafts PC All RESOLVED FIXED P2 normal --- 1 mike jonathan.robie liam public-qt-comments oldest_to_newest 115614 0 mike 2014-11-27 11:16:46 +0000 @@ comments on the text flagged with "@@" [76] UnaryLookup ::= "?" KeySpecifier [53] KeySpecifier ::= NCName | IntegerLiteral | ParenthesizedExpr | "*" UnaryLookup returns a sequence of values selected from the context item, which must be a map or array. If the context item is not a map or an array, an error is raised [err:XPTY0tbd]. @@ already raised: tbd needs to be defined. If the KeySpecifier is not a wildcard @@ nowhere is it said that "wildcard" means "*" , the semantics of the UnaryLookup operator are as follows: Let KS denote the items in the sequence to which the KeySpecifier evaluates. The UnaryLookup operator is equivalent to the following expression: for $k in KS return .($k) @@ This equivalence does not hold in the error case where teh context item is a function other than a map or array. Unary lookup is used primarily in predicates (e.g. $map[?name='Mike'] or with the simple mapping operator (e.g. $maps ! ?name='Mike'). @@ In XSLT, which makes much more use of the context item, it is likely to be used much more widely, e.g <xsl:sort select="?id"/>. See 3.11.3.2 Postfix Lookup for the postfix lookup operator. Examples: ?name is equivalent to .("name"), an appropriate lookup for a map. ?2 is equivalent to .(2), an appropriate lookup for an array or an integer-valued map. ?($a) is equivalent to for $k in $a return .($k), allowing keys for an array or map to be passed using a variable. ?(2 to 4) is equivalent to for $k in (2,3,4) return .($k), a convenient way to return a range of values from an array. @@ It might be worth pointing out that the function signature for array functions expects xs:integer, the supplied value is atomized, and converted to integer if it is untyped atomic. But supplying a double will give a type error. So if @id is untypedAtomic "3", then ?(@id) works, but ?(@id+1) does not. @@ It might also be worth a reminder that you get a dynamic error if the subscript is out of bounds. For example ([1,2,3], [1,2,5], [1,2])[?3 = 5] gives an error because ?3 on one of the items in the sequence fails. However, exists(([1,2,3], [1,2,5], [1,2])[?3 = 5]) might succeed because of early exit. If the KeySpecifier is "*" and the context item is a map, unary lookup is equivalent to the following expression: for $k in map:keys(.) return .($k) @@ We don't provide an equivalence for the case where the KeySpecifier is "*" and the context item is an array. I think the equivalence is for $k in 1 to array:size(.) return .($k) 116081 1 liam 2014-12-10 01:07:09 +0000 I think all of this is editorial execpt choosing which error gets raised; is that right? 116293 2 jonathan.robie 2014-12-15 01:13:37 +0000 (In reply to Michael Kay from comment #0) > @@ already raised: tbd needs to be defined. Done. > If the KeySpecifier is not a wildcard > > @@ nowhere is it said that "wildcard" means "*" > > , the semantics of the UnaryLookup operator are as follows: Let KS denote > the items in the sequence to which the KeySpecifier evaluates. The > UnaryLookup operator is equivalent to the following expression: > > for $k in KS > return .($k) > > @@ This equivalence does not hold in the error case where teh context item > is a function other than a map or array. The sentence before this says it's an error if the context item is not a map or array: <quote> UnaryLookup returns a sequence of values selected from the context item, which must be a map or array. If the context item is not a map or an array, a dynamic type error is raised [err:XPTY0004]. </quote> > Unary lookup is used primarily in predicates (e.g. $map[?name='Mike'] or > with the simple mapping operator (e.g. $maps ! ?name='Mike'). > > @@ In XSLT, which makes much more use of the context item, it is likely to > be used much more widely, e.g <xsl:sort select="?id"/>. I deleted the word "primarily. > See 3.11.3.2 Postfix Lookup for the postfix lookup operator. > > Examples: > > ?name is equivalent to .("name"), an appropriate lookup for a map. > ?2 is equivalent to .(2), an appropriate lookup for an array or an > integer-valued map. > ?($a) is equivalent to for $k in $a return .($k), allowing keys for an array > or map to be passed using a variable. > ?(2 to 4) is equivalent to for $k in (2,3,4) return .($k), a convenient way > to return a range of values from an array. > > @@ It might be worth pointing out that the function signature for array > functions expects xs:integer, the supplied value is atomized, and converted > to integer if it is untyped atomic. But supplying a double will give a type > error. So if @id is untypedAtomic "3", then ?(@id) works, but ?(@id+1) does > not. OK. I added these examples. > @@ It might also be worth a reminder that you get a dynamic error if the > subscript is out of bounds. For example > > ([1,2,3], [1,2,5], [1,2])[?3 = 5] gives an error because ?3 on one of the > items in the sequence fails. However, exists(([1,2,3], [1,2,5], [1,2])[?3 = > 5]) might succeed because of early exit. I added the first part of this but not the second. > @@ We don't provide an equivalence for the case where the KeySpecifier is > "*" and the context item is an array. I think the equivalence is > > for $k in 1 to array:size(.) > return .($k) Ouch! Back when arrays were maps, the old equivalence worked, but this did need fixing. Done now.