Re: Joint WG decisions affecting XSLT 3.0

> On 30 Sep 2016, at 10:36, Michael Kay <mike@saxonica.com> wrote:
> 
> 
> 
> Two decisions made on 27 Sept affect the XSLT spec:
> 
> (a) Bug 29872: brings parse-json() and json-to-xml() into line by making clear that when escape=true() is set in the options, existing escape sequences in the input are retained. For example if "%" is unnecessarily escaped in the input as \u0025, then it ends up in the output as \u0025. Previously parse-json() was specified this way but json-to-xml() left the question open.
> 

An update on this. I raised bug 29872 because of an inconsistency in the rules for parse-json(): there was ambiguity about whether existing escape sequences should be retained. I had assumed that the WG's intent was that they should be retained, and proposed (a) clarifying parse-json() to make this clear, and (b) bringing json-to-xml() into line.

Josh has now pointed out that the inconsistency was in fact the result of incomplete implementation of the changes agreed in July 2015 for bug 28812. The recorded decision on that bug was that existing escape sequences should NOT be retained, but I failed to spot and delete a rule saying the opposite.

So if the WGs confirm the original decision on bug 28812 then we will have to roll back the changes made by bug 29872.

Michael Kay
Saxonica

Received on Friday, 30 September 2016 17:50:53 UTC