Re: Bug in Mozilla parser

Vinu wrote:
> What i can see in memory(i use C/C++ version of the parser in
> VisualStudio) is that the XML buffer is represented as one byte per
> character.
> For e.g space char is represented as 32(dec) and so on.
> Now this buffer also contains the character ¤ which has ASCII code
> 164(decimal) or A4(hex).
> When the parser starts parsing this buffer to build a DOM tree, it is
> replacing ¤(code 164 and represented as one byte) with two bytes(which
> when represented in ASCII are the characters ¤). I think this is an
> UTF-8 representaion, where two bytes are used to represent ¤, whereas
> in ASCII/latin1 it would be represented with just one byte.

OK, now I understand. Sorry for being so slow. 

> So when i get back the XML from the DOM, and i am expecting a latin1
> encoding, i get ¤ wherever ¤ was expected.

Does your XML document start with an XML Declaration specifying
ISO-8859-1 as the character encoding? If not, then UTF-8 is assumed.

> So is the behaviour of any XML parser to store XML internally in UTF-8

Given that parsers are mandated to support UTF-8, probably yes.

> and then probably its the responsibility of the application using this
> parser to convert this XML in UTF-8 encoding to any encoding which it
> wishes.

Yes, that's probably true also.

///Peter
-- 
XML FAQ: http://xml.silmaril.ie/