Re: [Comment on xml-i18n-bp WD] CDATA sections

Hello Elliotte,

Just a personal comment for the moment:

I think the caution against CDATA sections is motivated by the fact
that you cannot use numeric character references in CDATA sections.
This means that unless you use an Unicode encoding for the document,
there are limitations on the set of characters you can use.
As a specific example, if you have an XSLT stylesheet with
<xsl:output encoding='iso-8859-1'
       cdata-section-elements='[lots of elements]' />
the risk is high that your stylesheet will fail because it can't
output things the way you specify (well, XSLT 1.0 has a should for
cdata-section-elements, so an XSLT processor can close the
CDATA section and reopen it afterwards).

There was a time when some people claimed that CDATA sections were
more than syntactic sugar, i.e. something like one more, somewhat
special, element. In such a kind of use, the i18n concerns were most
obvious. The fact that CDATA sections don't appear in the
Infoset and are flattened by XSLT has fortunately let this view
die out as far as I'm aware of.

Regards,    Martin.

At 19:13 06/05/23, Elliotte Harold wrote:
 >
 >I agree whole-heartedly with your sentiment "Use CDATA sections with 
caution." I am not aware, however, of internationalization related reasons 
for this rule. I'm curious to see how you fill out this section. Doubtless 
there are issues I haven't thought of here.
 >
 >--
 >         Elliotte Rusty Harold  elharo@metalab.unc.edu
 >XML in a Nutshell 3rd Edition Just Published!
 >http://www.cafeconleche.org/books/xian3/
 >http://www.amazon.com/exec/obidos/ISBN=0596007647/cafeaulaitA/ref=nosim
 >

Received on Tuesday, 23 May 2006 10:39:58 UTC