RE: Handling of list of strings having pattern facets in preserve.lexical mode

Dear Antoine,

Thank you for your question about the EXI Proposed Recommendation. When the
value of the preserve.lexical option is false, all values are represented
according to their associated schema datatypes. However, when the value of
the preserve.lexical option is true, it is the intent of the specification
that all values be represented as generic strings without regard for their
associated schema datatype. As such, the facets of the schema datatypes do
not come into consideration when that value of the preserve.lexical option
is true. Only the facets defined in table 7-2 come into play. 

On review, the specification could be more clear on this point. We will look
into clarifying this in a future version of the specification. Thank you for
calling our attention to this potential point of confusion.

	Best wishes,

	John

AgileDelta, Inc.
john.schneider@agiledelta.com
http://www.agiledelta.com
 
 
 

> -----Original Message-----
> From: public-exi-comments-request@w3.org 
> [mailto:public-exi-comments-request@w3.org] On Behalf Of 
> Antoine Mensch
> Sent: Thursday, February 10, 2011 1:28 AM
> To: public-exi-comments@w3.org
> Subject: Handling of list of strings having pattern facets in 
> preserve.lexical mode
> 
> Hi,
> 
> We have a question regarding the handling of list of strings 
> having pattern facets in preserve.lexical mode: while this is 
> a "cornercase", the spec is not completely clear about how it 
> should be handled:
> * In preserve.lexical mode, lists must be encoded as Strings 
> (section 7).
> * The restricted character set of a List type is the 
> restricted character set of its item type (section 7).
> * There is no default character set to be used in 
> preserve.lexical for the String Datatype: so we assume that 
> the restricted character set to be used for a XSD type 
> derived from xsd:string is the one defined based on the 
> pattern facets, as described in section 7.1.10.1.
> 
> If we follow the above rules, and the restricted character 
> set derived from the patterns does not contain whitespaces 
> (which is likely, as whitespaces are not welcome in string 
> values used in lists), then we will have a deviation from the 
> restricted character set and an explicit codepoint for each 
> whitespace in the list lexical representation.
> 
> Is that the correct interpretation and the intent of the 
> spec? A simple fix could be to add whitepaces to the 
> restricted character set of lists when deriving it from the 
> item type restricted character set, but that would require a 
> change of the spec.
> 
> Cheers
> 
> Antoine Mensch
>

Received on Wednesday, 2 March 2011 17:49:03 UTC