Bug 4471 – [FO] Casting date/time timezone offset 0 values to string/untypedAtomic

Bug 4471 - [FO] Casting date/time timezone offset 0 values to string/untypedAtomic

Summary:

[FO] Casting date/time timezone offset 0 values to string/untypedAtomic

Status:	CLOSED FIXED

Product:	XPath / XQuery / XSLT
Component:	Functions and Operators 1.0
Version:	Recommendation
Platform:	PC Windows XP

Importance:	P1 normal
Target Milestone:	---
Assigned To:	Michael Kay
QA Contact:	Mailing list for public feedback on specs from XSL and XML Query WGs

URL:
Whiteboard:
Keywords:

Depends on:
Blocks:
	Show dependency tree / graph

Reported:	2007-04-18 19:07 UTC by Michael Rys
Modified:	2007-11-16 09:14 UTC (History)
CC List:	0 users (show)

See Also:

Attachments
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.

Description Michael Rys 2007-04-18 19:07:24 UTC

The current casting rules for values of date/time type with timezone offset 0
values to string/untypedAtomic generate the serialization +00:00 instead of the
shorter and much more often used Z. Note that all the examples in the document
normally use Z instead of +00:00 and a few existing implementations also use Z
(Saxon, SQL Server).

We therefore request that the casting rules get changed to produce a Z instead.

Searches in the old documents appear to have moved from the rule (last seen in
http://www.w3.org/TR/2005/WD-xpath-functions-20050211/)

<cite>If necessary, the localized value is recovered from the normalized value
as follows: if an explicit timezone was present, the normalized value is
adjusted using the explicit timezone; if an explicit timezone was not present,
the Z timezone is dropped from the normalized value. The localized value and
the explicit timezone, if present, are cast separately to xs:string and
concatenated to yield TV.</cite>

which seem to imply to work with Z, to the current rule... since this was an
editorial rework and after talking to the editors, we assume that this was an
oversight.

This will also affect the data model serialization.

Comment 1 Michael Kay 2007-04-23 21:37:30 UTC

Note also, although QT has departed from the definitions in XML Schema as
regards timezone preservation, the canonical representation in XML Schema does
denote UTC by "Z", and there is no reason why we should choose to differ from
this.

Comment 2 Michael Kay 2007-05-01 21:46:37 UTC

Proposed change: in 17.1.5 Casting to Date and Time Types, change

declare function eg:convertTZtoString($tz as xs:dayTimeDuration?) as xs:string
{
   if (empty($tz)) then ""
   else 
     let $tzh := fn:hours-from-dayTimeDuration($tz)
     let $tzm := fn:minutes-from-dayTimeDuration($tz)
     let $plusMinus := if ($tzh >= 0) then "+" else "-"
     let $tzhString := eg:convertTo2CharString(fn:abs($tzh))
     let $tzmString := eg:convertTo2CharString(fn:abs($tzm))
     return fn:concat($plusMinus, $tzhString, ":", $tzmString)
}

to

declare function eg:convertTZtoString($tz as xs:dayTimeDuration?) as xs:string
{
   if (empty($tz)) 
     then ""
   else if ($tz eq xs:dayTimeDuration('PT0H'))
     then "Z"
   else 
     let $tzh := fn:hours-from-dayTimeDuration($tz)
     let $tzm := fn:minutes-from-dayTimeDuration($tz)
     let $plusMinus := if ($tzh >= 0) then "+" else "-"
     let $tzhString := eg:convertTo2CharString(fn:abs($tzh))
     let $tzmString := eg:convertTo2CharString(fn:abs($tzm))
     return fn:concat($plusMinus, $tzhString, ":", $tzmString)
}

Comment 3 Michael Kay 2007-05-18 10:32:06 UTC

The proposed change was accepted in the joint meeting 329 on 2007-05-08, with
one minor change: the zero-length duration should be expressed in its canonical
form as PT0S.

The change will appear as erratum FO.E6

Comment 4 Michael Kay 2007-06-12 20:17:25 UTC

I am reopening this bug because of an unrelated error in the same function.
Both the original and the "corrected" versions of the internal function
eg:convertTZtoString use the functions fn:hours-from-dayTimeDuration and
fn:minutes-from-dayTimeDuration, which no longer exist. The functions are now
called fn:hours-from-duration and fn:minutes-from-duration.

The collective eyeballs of the WG seem to have been out of focus...

Comment 5 Michael Kay 2007-06-27 14:54:55 UTC

The WGs approved this change on 27 June 2007