Re: Using CDATA sections for element content??

[Eric Bohlman <ebohlman@earthlink.net>]

2/7/01 11:23:11 AM, Jerry Murray <Jmurray@Ironplanet.com> wrote:

>A business partner has asked that we use CDATA sections for all element
>content so they don't have to worry about handling special characters.  
>
>I'd rather not since it handling special characters seems like a trivial
>problem.  There is a clear issue of whether the "raw" document is easier to
>"read" if it uses " ...<![CDATA[ ..." as opppose to using escape sequences.
>But for our application, the "raw" document will very seldom ever be view in
>the text form so these distinctions are not very important.  This is also a
>very limited application with basically simple content.  We are not using
>entities or namespaces in the content so the issues associated with those
>are not a factor.
>Are there other issues that I have not identified?

This requirement makes me suspicious that your partner isn't using a real XML 
parser on your output but instead is hack-parsing it with Perl regexps or the 
like.  In that case, there are probably a bunch of hidden dependencies on 
things like character encodings, use of numeric character references, and even 
internal tag structure (a lot of hack attempts can't cope with perfectly legal 
things like embedded newlines in tags, and some make assumptions about what 
tags go on what lines).  IOW, there are likely to be a lot of dependencies on 
purely lexical details that shouldn't really matter.