In character data, > isn't reserved, but ]]> is, so it's worth escaping
> anyway rather than only looking for those occurrences of > that are
preceded by ]]. Why is ]]> reserved when you're not in a CDATA section?
I don't know - arguably for robustness, in case the CDATA opener is
mistyped. Or perhaps it's some SGML legacy I'm not aware of (there's a
lot of it around).
In attribute values, > is reserved, which is pretty pointless, and this
is entirely due to SGML legacy, which allows the quotes around the
attribute value to be omitted.
Don't look for too much logic here. It's like asking why "a" is where it
is on the keyboard. There's a historical explanation, but not a rational
decision-making process.
Michael Kay
Saxonica
