4550 2007-05-10 03:50:39 +0000 Is an assertion required for {http location} EBNF grammar? 2007-05-10 15:45:24 +0000 1 1 1 Unclassified WSDL Adjuncts 2.0 All All RESOLVED FIXED http://lists.w3.org/Archives/Public/www-ws-desc/2007May/0015.html resolved P2 normal --- 1 jonathan plh www-ws-desc oldest_to_newest 15062 0 jonathan 2007-05-10 03:50:39 +0000 I'd like to propose replacing the text associated with the EBNF grammar in section 6.8.1.1: "The following EBNF [ISO/IEC 14977:1966] grammar represents the patterns for constructing the request IRI" with the new assertion HTTPSerialization-2106: "The {http location} property MUST conform to the following EBNF [ISO/IEC 14977:1966] grammar, which represents the patterns for constructing the request IRI." regards, John Kaputin. On 5/4/07, John Kaputin (gmail) <jakaputin@gmail.com> wrote: > > There is no WSDL assertion stating that an {http location} or > whttp:location must conform to the EBNF grammar defined in Part 2 Section > 6.8.1.1. > > Consider the following whttp:location values, which do not conform to this > EBNF grammar: > > "/to}wn/{localname}" (unmatched left brace) > "/town/{local:name}" (template does not specify an NCName) > "/town/{localname" (closing right brace is missing) > > Should these be considered WSDL errors? If so, should there be an > assertion about the EBNF grammar? Apache Woden parses the whttp:location > attribute and if it does not conform to the EBNF grammar the {http location} > property is flagged as invalid, but there is no WSDL assertion to use for > error reporting. > > Alternatively, Woden could just ignore these grammar errors and treat them > as ordinary string content in {http location} but the problem would still > need to be resolved when the request IRI is constructed by the message > builder. 15063 1 jonathan 2007-05-10 15:15:34 +0000 Proposal accepted.