Re: Format of protocol error bodies

On 19/06/12 15:36, Lee Feigenbaum wrote:
> One of our open comments is RC-2:
>
> http://lists.w3.org/Archives/Public/public-rdf-dawg-comments/2010Sep/0002.html
>
>
> To summarize, this was a comment from before we moved the SPARQL
> Protocol away from WSDL. In this comment, Richard Cyganiak was asking
> that we consider standardizing the format of fault message bodies -- he
> suggested at the time using the XML serialization based on the WSDL for
> malformed queries and query request refused.
>
> His message sparked a long discussion thread, that included proponents
> of plain text error reporting, of content negotiation, and of leaving
> the format of error message bodies underspecified.
>
> There was no consensus.
>
> We as a working group have not spent time on the issue, but ought to
> discuss it via email or on a call, before replying to the message.
>
> Greg and I have discussed this and believe that the format of error
> message bodies should remain underspecified,

+1

> primarily due to:
>
> * A lack of significant implementer experience with one particular
> format of error bodies
> * A lack of clear community consensus

* error pages generated by web containers may not easily be under 
application control, being shared amongst contained apps

* organisations like to present their image carefully on errors - c.f. 
github's 404 page or twitter which has a search bar.

> We also think that better HTTP status messages would help, and we'll
> update the examples in protocol to illustrate this. (Though this won't
> be any sort of conformance requirement.)
>
> What do you think?

I am not convinced there is a single, "right" answer.

>
> Lee
>

 Andy

Received on Tuesday, 19 June 2012 14:44:32 UTC