RE: white space in parse attribute

At 10:38 AM -0800 1/7/04, Jonathan Marsh wrote:
>The XML Core WG was reluctant to add additional processing requirements
>(even the modest one of stripping whitespace) for this case.  We
>collectively couldn't recall in practice seeing whitespace around names
>in similar specs, and didn't think the overhead was necessary.
>Whitespace in the parse attribute will cause the XInclude processor to
>fail, and users must refrain from putting it in.  The note in section
>3.1 covers the case where this error is not detected by certain
>processors because the whitespace has been normalized away prior to
>inclusion.
>
>Please let us know if you disagree with this resolution.
>

OK. I understand your reasoning and it makes sense. To that extent I 
accept your resolution. However, based on that logic, I would like to 
make a couple of related comments.

1. Why is the parse attribute declared to be an enumeration? Why not 
declare it to be CDATA so that leading and trailing white space isn't 
stripped? This would better guarantee interoperability.

Of course the DTD is non-normative, and people can declare it to be 
whatever they want. Since users can use any declaration they want 
(including in the current case of an enumeration) there is non extra 
work for the processor here. It still has to explicitly check that 
the value is either text or xml.

I suppose this isn't a substantive change, merely an editorial one. 
Change the non-normative DTD fragment in 3.1 to use CDATA rather than 
an enumeration so users are less likely to be tripped up on this 
point.

2. (Editorial) Section 3.1 defines the parse attribute as "An 
enumeration specifying whether to include the resource as parsed XML 
or as text." I am not sure what it really mean in this context for it 
to be an "enumeration". I'm not sure that term really has meaning 
when applied to the instance document as opposed to the DTD. At most, 
it's an attribute of type enumeration. However, in practice very few 
modern parsers call out enumerated attributes as a different type 
than CDATA. (Some older parsers did this.)

I suggest removing the word "enumeration" here and rewriting as 
something like "Indicates whether to include the resource as parsed 
XML or as text."
-- 

   Elliotte Rusty Harold
   elharo@metalab.unc.edu
   Effective XML (Addison-Wesley, 2003)
   http://www.cafeconleche.org/books/effectivexml            
   http://www.amazon.com/exec/obidos/ISBN%3D0321150406/ref%3Dnosim/cafeaulaitA

Received on Wednesday, 7 January 2004 15:27:09 UTC