W3C home > Mailing lists > Public > public-rdf-dawg-comments@w3.org > October 2010

Re: SPARQL 1.1 Protocol: Format of fault messages

From: Richard Cyganiak <richard@cyganiak.de>
Date: Sun, 3 Oct 2010 16:31:47 +0100
Cc: public-rdf-dawg-comments@w3.org
Message-Id: <BE04702F-3785-4D90-B3CC-0FF50B2BE18B@cyganiak.de>
To: Eric Prud'hommeaux <eric@w3.org>

On 1 Oct 2010, at 21:24, Eric Prud'hommeaux wrote:
>> 1. To state in the HTTP binding that clients SHOULD use the XML
>> fault message format when reporting faults.
>
> It may be polite to still have two, one for SOAP interfaces, which
> have some standard tooling and contracts, etc.

I'm just concerned about the HTTP bindings. I don't really see the  
SOAP bindings used out there in the wild; my clients certainly don't  
support them, and I had no user requests for SOAP support.

> I expect your principle
> issue is with the free-form beat poetry style of the rest bindings.

My issue is with two things:

1. The current spec says: “Use this HTTP status code, everything else  
is up to you.”

It should say: “You SHOULD report the error message this way.”

2. The current spec has an example that shows the error message  
embedded in HTML in a way that it can't be gotten out by a client.

It should not use such an example.

> We'd kind of like to leave room for implementors to innovate

Saying that you SHOULD use a certain format leaves enough wiggle room  
for implementors to innovate, IMO.

I think the spec should really communicate one no-brainer way of  
implementing error reporting, for anyone who really just wants to be  
conformant without any desire to do fancy innovative stuff. I really  
want that format to be machine-readable; beyond that I really don't  
care much.

I have no objection against also proposing alternative response  
formats, as long as one option is clearly highlighted as the “no- 
brainer” option.

> and reply with
> our baseline error response plus some structured stuff to say what
> line and character caused what sort of error.

I'm all for reporting line and column in a structured format if  
possible, but that would be the cherry on the cake.

I'd be sufficiently happy with just the cake.

> We'd also like that to
> show up in browsers. What if we say that it must appear in a pre with
> a particular class?
>
> <?xml version="1.0" encoding="UTF-8"?>
> <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0//EN" "http://www.w3.org/MarkUp/DTD/xhtml.dtd 
> ">
> <html xmlns="http://www.w3.org/1999/xhtml">
>  <head>
>    <title>D'oh!</title>
>  </head>
>  <body>
>    <pre class="malformedQueryFault" style="display:none">Parse error  
> at [5:17], unexpected ' '</pre>

Why would you make it display:none? If you use HTML already, then at  
least make the error message show up in the browser!

And RDFa would be better than a magic class name.

Best,
Richard
Received on Sunday, 3 October 2010 15:38:53 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Sunday, 3 October 2010 15:38:53 GMT