RE: [xml-dev] Attributes as tags in namespaces and how to guess character encoding

> 
> On the contrary, the xml declaration is entirely in ascii 
> except for a 
> possible byte order mark, so the processor can determine 8-bit vs. 
> 16-bit encodings from the BOM and the <?xml, and then read 
> the encoding 
> declaration, knowing that it is in ascii.

It's perfectly legitimate to use encodings that are not supersets of
US-ASCII, for example EBCDIC. A parser that wants to handle such encodings
has to examine the first few bytes in the file and see whether they match
"<?xml" in any of the encodings that it supports.

Michael Kay