Re: XSLT/XPATH problem, possible bug

Michael Hansen wrote:
>>My reading of <http://www.w3.org/TR/REC-xml/#dt-chardata> is that you
>>do not need to escape the ampersand '&' in a processing instruction:
>>
>>"The ampersand character (&) and the left angle bracket (<) MUST NOT
>>appear in their literal form, _except_ when used as markup delimiters,
>>or within a comment, a processing instruction, or a CDATA section."
>>
>>So you would simply need to write your xml-stylesheet processing
>>instruction with unescaped ampersands (e.g.
>>  &filter=Rock
>>) and all is fine, Mozilla will then make the proper URL request and
>>your PHP script will have the proper $_GET variables.
>>
>>That is what the XML specification gives.
> 
> 
> Aha, so strictly speaking it is not a bug. However it would make this a 
> great deal easier if encoded ampersands were allowed here too (in terms of 
> workflow).
> 
> Thanks for clearing this up,

Things are more complex, the specification xml-stylesheet, 
<http://www.w3.org/TR/xml-stylesheet/>, defines xml-stylesheet as a 
processing instruction but then makes further syntax restrictions on the 
data of the processing instruction where
<http://www.w3.org/TR/xml-stylesheet/#The%20xml-stylesheet%20processing%20instruction>
defines pseudo attributes and their syntax as

[1]    	StyleSheetPI 	   ::=    	'<?xml-stylesheet' (S PseudoAtt)* S? '?>' 	
[2]    	PseudoAtt 	   ::=    	Name S? '=' S? PseudoAttValue 	
[3]    	PseudoAttValue 	   ::=    	('"' ([^"<&] | CharRef | 
PredefEntityRef)* '"' 	
			| "'" ([^'<&] | CharRef | PredefEntityRef)* "'") 	
			- (Char* '?>' Char*) 	
[4]    	PredefEntityRef 	   ::=    	'&amp;' | '&lt;' | '&gt;' | '&quot;' 
| '&apos;'

which means reading the XML specification does not suffice, the 
xml-stylesheet specification makes e.g.
   href="someURI"
a pseudo attribute value where in the attribute value any '<' or '&' 
unescaped is disallowed while '&lt;' and '&amp;' are allowed and 
necessary it seems.

So with that specification it seems any processor making an HTTP request 
on the URI in the href pseudo attribute should unescape any '&amp;' as 
'&' as the value of the pseudo attribute is "the value after each 
reference is replaced by the character it references".

It looks like Mozilla has a bug after all, I might file one tomorrow if 
noone shows up in the meantime opposing to the above reasoning.


-- 

	Martin Honnen
	http://JavaScript.FAQTs.com/