Re: RFC822 mail in CDATA section

[Joseph Kesselman <ke...@us.ibm.com>]

The answer is that you can't count on using a single CDATA section; you 
have to be prepared to either exit the CDATA section for a moment so part 
or all of the ]]> text is in normal text or splitting it across successive 
CDATA sections.
        <![CDATA[This works: ]]>]]><![CDATA[ -- note that > generally 
doesn't need to be escaped.]]>
        <![CDATA[So does this: ]]]]><![CDATA[> -- or this 
]]]><![CDATA[]>]]>
Obviously your XML generator has to recognize this case and deal with it, 
and when you parse the document again you have to accept that your text 
may be split across multiple text nodes/events.

Note that a similar hassle arises if any of the characters in the data 
can't be expressed directly in the encoding you've chosen for your XML 
file. CDATA sections *only* support that encoding; if you need to do a 
numeric character reference, you need to exit the CDATA section... or 
re-encode all the data in something like base-64, in which case the CDATA 
section really isn't helping you.


The better solution, in most cases, is to just write out text, escaping 
those characters which need to be escaped. CDATA sections are really 
intended as a convenience for humans hand-editing XML files without the 
benefit of XML-aware tools... and as XML knowledge/support becomes 
ubiquitous and people push the boundaries of this trivial workaround, they 
start to become more annoying than useful. Unless you are working with an 
archaic tool that doesn't understand numeric character escapes, I would 
recommend *not* using CDATA sections. As silver bullets go, they're very 
tarnished and rather alloyed.

______________________________________
Joe Kesselman  / IBM Research


---------------------------------------------------------------------
To unsubscribe, e-mail: xerces-j-user-unsubscribe@xml.apache.org
For additional commands, e-mail: xerces-j-user-help@xml.apache.org