On Tue, 03 Apr 2001, you wrote:
> We just got a shock. We've been embedding stuff as <CDATA> expecting that
> it would be passed through whatever transforms (I believe JSP, in our
> case) untouched. But, apparently that isn't the case. CDATA is parsed at
> least with regards to line endings and extra spaces by our parser, or
> at least, by the version of our parser that is running under Solaris. Is
> that supposed to happen?
No. If it does, your processor is buggy.
(I'm assuming you do mean CDATA Marked Sections, not something
> Is there some alternative way to encode text
> (the equivalent of the HTML <PRE> that says, "leave this alone, don't
> parse it"?
That's what a CDATA Marked Section is for.
> I thought this was precisely the difference between CDATA and PCDATA--
> that the latter, but only the latter was parsed.
Broadly correct. However, the internal curiosities of SGML, on
which XML is founded, do leave open some questions of how to
handle RS and RE (record-start and record-end), so I can imagine
someone could make an argument for doing what you're complaining
about and claiming it's valid :-(