Re: CR117 Option 1

Jonathan, is it your intention here that an encoded parameter in the
path can still contain multiple path segments ('/') or even start the
query string ('?')?

Shouldn't / and ? be unencoded in the query, and encoded in the path?
Indeed, RFC3986 allows / and ? to be unencoded in the query. This would
be opposite to what you have.

Jacek

On Wed, 2007-02-07 at 13:57 -0800, Jonathan Marsh wrote:
> Remove the text in 6.7.1.1 that reads:
> 
>  
> 
> No percent-encoding mechanism, as defined in section 2.1 of RFC 3986,
> is performed on the replacement value.
> 
>  
> 
> Add to the bullet list in 6.7.1.1:
> 
>  
> 
> Ā·        A template whose element name is preceded by the character
> ā€œ~ā€ is replaced by the corresponding XML element value without
> additional %-encoding performed.  Otherwise: 
> 
>  
> 
> A template preceded in the {http location} by a ā€œ?ā€ or ā€œ#ā€ character,
> or an uncited parameter, is encoded as follows:  Any character in the
> XML value falling outside the range ALPHA | DIGIT | "-" | "." | "_" |
> "~" | "!" | "$" | "&" | "'" | "(" | ")" | "*" | "+" | "," | ";" | "="
> | ā€œ:ā€ | ā€œ@ā€ MUST be %-encoded.  
> 
>  
> 
> A template not preceded in the {http location} by a ā€œ?ā€ or ā€œ#ā€
> character is encoded as follows:  Any character in the XML value
> falling outside the range ALPHA | DIGIT | "-" | "." | "_" | "~" | "!"
> | "$" | "&" | "'" | "(" | ")" | "*" | "+" | "," | ";" | "=" | ā€œ:ā€ |
> ā€œ@ā€ | ā€œ?ā€ | ā€œ/ā€ MUST be %-encoded.
> 
>  
> 
> %-encoding MUST be performed using the UTF-8 representation of the
> character as prescribed by section 6.4 of [IRIs].
> 
>  
> 
> Add to the bullet list in 6.7.2.2.1:
> 
> Ā·        Parameter values falling outside the range ALPHA | DIGIT |
> "-" | "." | "_" | "~" | "!" | "$" | "&" | "'" | "(" | ")" | "*" | "+"
> | "," | ";" | "=" | ā€œ:ā€ | ā€œ@ā€ MUST be %-encoded.  %-encoding MUST be
> performed using the UTF-8 representation of the character as
> prescribed by section 6.4 of [IRIs].
> 
>  
> 
> Change the BNF (wherever we decided to stick it) to:
> 
>  
> 
> httpLocation ::= CharData? (( openBrace | closeBrace | elementName )
> CharData?)*
> 
>  
> 
> CharData ::= [^{}]*
> 
>  
> 
> openBrace ::= '{{'
> 
>  
> 
> closeBrace ::= '}}'
> 
>  
> 
> elementName ::= rawElementName | encodedElementName
> 
>  
> 
> rawElementName ::= '{~' NCName '}'
> 
>  
> 
> encodedElementName ::= '{' NCName '}'
> 
>  
> 
>  
> 
>  
> 
> Notes on this proposal:
> 
>  
> 
> 1) This will no doubt require editorial license, For instance, see
> Philippeā€™s proposal for changed text for 6.7.1.1 at
> http://lists.w3.org/Archives/Public/www-ws-desc/2006Sep/0034.
> 
>  
> 
> 2) From rfc3986, within the query part or fragment the following
> characters may appear without %-escaping:
> 
>  
> 
> ALPHA | DIGIT | "-" | "." | "_" | "~" | "!" | "$" | "&" | "'" | "(" |
> ")" | "*" | "+" | "," | ";" | "=" | ā€œ:ā€ | ā€œ@ā€ | ā€œ/ā€ | ā€œ?ā€
> 
>  
> 
> 3) From rfc3986, Within a path part the following characters may
> appear without %-escaping:
> 
>  
> 
> ALPHA | DIGIT | "-" | "." | "_" | "~" | "!" | "$" | "&" | "'" | "(" |
> ")" | "*" | "+" | "," | ";" | "=" | ā€œ:ā€ | ā€œ@ā€
> 
>  
> 
> 4) Over-encoding of parameter values (encoding ā€œ/ā€ and ā€œ?ā€ even though
> itā€™s not necessary) is a possible simplification.
> 
>  
> 
> 5) I donā€™t think it makes sense escaping something other than a path
> part, query parameter, or fragment, so Iā€™m not considering specialized
> escaping in those places.
> 
>  
> 
> 6) I believe that the encoding, and determination of whether the
> template appears in the query parameter/fragment or somewhere else,
> can be made independently of resolving the reference against the
> {address}.  Someone might want to double check this belief.
> 
>  
> 
>  
> 
> Jonathan Marsh - http://www.wso2.com -
> http://auburnmarshes.spaces.live.com
> 
>  
> 
>  
> 
>