24200 2014-01-03 13:08:35 +0000 '"as" SequenceType' vs 'TypeDeclaration' in XPath/XQuery 3.0? 2014-01-07 18:36:27 +0000 1 1 1 Unclassified XPath / XQuery / XSLT XPath 3.0 Proposed Recommendation All All RESOLVED WONTFIX P2 trivial --- 1 tgraham jonathan.robie jmdyck mike public-qt-comments oldest_to_newest 97962 0 tgraham 2014-01-03 13:08:35 +0000 Is it an omission that in XPath 3.0 [1], '"as" SequenceType' appears in production [64] whereas 'TypeDeclaration' appears in production [3], where 'TypeDeclaration' is defined as '"as" SequenceType' in production [66]? [64] InlineFunctionExpr ::= "function" "(" ParamList? ")" ("as" SequenceType)? FunctionBody [2] ParamList ::= Param ("," Param)* [3] Param ::= "$" EQName TypeDeclaration? [66] TypeDeclaration ::= "as" SequenceType XQuery 3.0 [2] similarly uses '"as" SequenceType' in its production [165] while making rather more uses of 'TypeDeclaration' in other productions. [1] http://www.w3.org/TR/xpath-30/ [2] http://www.w3.org/TR/xquery-30/#id-inline-func [3] http://www.w3.org/TR/xpath-30/#prod-xpath30-Param [64] http://www.w3.org/TR/xpath-30/#prod-xpath30-InlineFunctionExpr [66] http://www.w3.org/TR/xpath-30/#prod-xpath30-TypeDeclaration [165] http://www.w3.org/TR/xquery-30/#prod-xquery30-InlineFunctionExpr 98090 1 jonathan.robie 2014-01-07 16:39:39 +0000 This is editorial. While I agree that it would be nice to make the two specs editorially consistent in the names of productions, the grammars are equivalent, and I would rather not change the grammar in a PR. I propose to fix this in 3.1. 98092 2 mike 2014-01-07 17:05:06 +0000 (Jonathan perhaps failed to make it clear that his response is given after WG discussion and decision. The machinery for generating the grammars used in our specs is quite complex and refactoring can have unexpected consequences, so we felt that this change was too high risk to make at this stage of the game). 98098 3 jmdyck 2014-01-07 18:36:27 +0000 It looks like we use 'TypeDeclaration' when specifying the type of a variable or parameter, and 'as SequenceType' when specifying the return type of a function or function-type. I think it would look somewhat odd (possibly confusing) to replace 'as SequenceType' with 'TypeDeclaration' in the latter productions. So, to answer the initial question, I suspect it's not an omission (or oversight) that 'TypeDeclaration' isn't used everywhere it could be. And I'm inclined to think it doesn't need "fixing" in 3.1.