Re: Attribute normalization

Hi Kevin,
[...]
> So, it seems that only #x20 characters will be seen in attribute values.
> Why does the
> spec mention the other values (#xD, #xA, #x9)?


Because as said in the first point of 3.3.3
"--a character reference is processed by appending the referenced character
to the attribute value"
if you had 
 in an attribute value of your inital document, then the
processor would have dereferenced the character to #xD so when you serialize
it, you have a #xD character in atribute value.

David