Re: relaxed and w3c validation service

Hello Jirka,

Thanks for your quick reply. As discussed earlier, I am copying the  
qa-dev list in my reply, in order to include the developer(s) of  
unicorn in the discussion.

On Jul 29, 2006, at 00:37 , Jirka Kosek wrote:
> OK, it should be fairly easy to map our current XML output format  
> to this one. Is there any interface specification for request? (I  
> was not able to find it on UniCORN pages.)
>
> Or is it expected that our response will be sent back after normal  
> HTTP GET request with several parameters which can be different for  
> each observer? Of course, we are preferring simple GET request over  
> full SOAP-based web-service.

It's the latter. Each observer can define its own request parameters,  
but they need to be documented in a specific format, consisting of a  
wadl (Web Application Description Language) file and an rdf file. The  
format is rather simple, and I am attaching a sample to this mail.  
The documentation for these files is at:
http://www.w3.org/QA/2006/obs_framework/?cat=contract

> I think that it will be useful if location of error can be  
> specified not only by line/column but also by XPath expression (if  
> document being validated is SGML or XML instance). Currently  
> Relaxed doesn't support this, but we probably will implement this  
> in a future because line/column paradigm is not very robust for  
> compound documents that have to be separated before validation.

That's a good point, I think this is something we had been discussing  
internally too. (yes, found it, see e.g [1] for a writeup on the  
question) If I am not mistaken however, XPath may be pretty useless  
if the tested document is very broken (not well formed, etc), so even  
if XPAth is a good idea, it can't be used in every situation either,  
can it?

[1] http://esw.w3.org/topic/MarkupValidator/M12N


-- 
olivier