29989 2016-11-07 06:52:08 +0000 [XQ31] CurlyArrayConstructor does not use EnclosedExpr 2016-11-11 14:32:58 +0000 1 1 1 Unclassified XPath / XQuery / XSLT XQuery 3.1 Candidate Recommendation PC Linux RESOLVED FIXED P2 normal --- 1 msclrhd jonathan.robie andrew_coleman jmdyck josh.spiegel public-qt-comments oldest_to_newest 128093 0 msclrhd 2016-11-07 06:52:08 +0000 The XQuery 3.1 EBNF grammar contains: [176] CurlyArrayConstructor ::= "array" "{" Expr? "}" However, other constructs of this form have been converted to use EnclosedExpr. This construct should be converted to: [176] CurlyArrayConstructor ::= "array" EnclosedExpr Unless there was a reason for not using EnclosedExpr. 128098 1 jmdyck 2016-11-07 17:53:12 +0000 Related history: At meeting #621: https://lists.w3.org/Archives/Public/public-xsl-query/2015Oct/0071.html I was given action item 621-03 "to update the grammar for empty expressions in braces" (as outlined in the accompanying decision). In this message: https://lists.w3.org/Archives/Public/public-xsl-query/2015Nov/0001.html I reported what I'd done to complete the action item, and (at the very bottom) said: Also, in CurlyArrayConstructor I did NOT change "{" Expr? "}" because it wasn't in the list of expressions to 'consider', despite that it presumably has the "{} means {()}" semantics. (I say "presumably" because I don't think the spec actually defines the meaning of "array {}".) My message came up at meeting #622: https://lists.w3.org/Archives/Public/public-xsl-query/2015Nov/0011.html but the minutes don't say (and I don't recall) whether we considered the point about CurlyArrayConstructor. (It looks like the quote above was the last time that CurlyArrayConstructor was mentioned in public-xsl-query.) 128101 2 josh.spiegel 2016-11-07 18:40:55 +0000 So, there are really two independent problems: (1) Use EnclosedExpr in the CurlyArrayConstructor production (editorial?) (2) Make explicit that array{} is the same as array{()} 128102 3 josh.spiegel 2016-11-07 18:49:24 +0000 Actually, doing (1) takes care of (2). 128154 4 andrew_coleman 2016-11-11 13:56:12 +0000 The WG decided to adopt the resolution described in comment #2. Action A-660-08 raised to track this. 128157 5 jmdyck 2016-11-11 14:32:58 +0000 I have change the grammars for XPath 3.1 and XQuery 3.1 as suggested in comment #0.