http://norman.walsh.name/2003/08/24/cdata 6 75 2003-08-24 $Date: 2005-09-11 10:27:02 -0400 (Sun, 11 Sep 2005) $ Norman Walsh 2003 Norman Walsh A short, cheeky complaint about the sad state of XML books. Dale Carnegie Any fool can crticize, condemn, and complain—and most fools do. This essay is a little bit cheeky on my part. Authors work hard, I know from personal experience, and maybe it's unfair to judge a book by a single section, to paraphrase a common saying. We all make mistakes, right? I'll try to ameliorate my criticism by not mentioning any names. But I have to ask, what is so hard to understand about CDATA sections? Wandering through the book store yesterday, I paused to skim the XML books. A few of years ago, just before the bubble burst, when the XML books were coming out fast and furious, I happened to notice an error in one of them. Something to do with CDATA sections, but I've forgotten exactly what, so now when I pick up a random XML book, I always look to see how CDATA sections are described. The first book I skim says that CDATA sections begin with <![cdata[ . Bzzzt. Wrong, but thank you for playing. The next book I pick up says that CDATA sections cannot contain ]]. Sigh. Nope. The third book gets it right. The fourth book correctly points out that CDATA sections cannot contain ]]> , but fails to mention that the string ]]> (unescaped) can't ever occur in a document except when it ends a CDATA section. Close, but no cigar. The last book I picked up didn't mention CDATA sections at all. No points off, I suppose. If I promised to teach you XML in four and a third minutes or something like that, maybe I wouldn't try to explain CDATA sections either. I wonder if XML 2.0 should even contain CDATA sections?