RE: New issue: HTTP binding/status code

Jean Jacques,

Since I put the "??" strings there... I'll fess up.

They are there because some judegment needs to be applied :-) 

They are certainly *NOT* intended to suggest that status code will do. 

They are there as a request for input - "What should these codes be?" The
ednote that follows [1] also indicates that the intent was to use status
codes in manner that was consistent with the resolution of issue 12 (which
was resolved against a very different narrative).

The second table is intended to enumerate all top level SOAP Faults and the
HTTP status codes to be used when transferring such a fault in an HTTP
response. 

Incidentally, I think with these tables in place... the "when is a fault a
fault" question that Mark Baker asks does not arise... the binding requires
that status codes consistent with the fault being carried are used and the
quoted fault use-case is *not* supported... without further encapsulation of
the quoted fault within the message (ie. a fault carried in a 200 is an
implementation error - a binding that did this would fail a conformance
test).

Stuart

> -----Original Message-----
> From: Jean-Jacques Moreau [mailto:moreau@crf.canon.fr]
> Sent: 22 March 2002 13:43
> To: xml-dist-app@w3.org
> Subject: New issue: HTTP binding/status code
> 
> 
> I would like to raise the following issue: two tables[1,2] in the
> HTTP binding contain the strings "??" and "???". The meaning of
> these characters is ambiguous: it could either mean "any status
> code is valid", or "we haven't though about this problem yet;
> work in progress". Also, it is not clear whether only a certain
> subset of the HTTP status code is acceptable, instead of all
> possible HTTP status code.
> 
> Jean-Jacques.
> 
> [1] 7.4.1.2.1 "Receiving State" table
> [2] 7.4.1.2.3 "Responding State SOAP Faults" table
> 
> 
> 

Received on Friday, 22 March 2002 10:47:22 UTC