Re: Uniform access to descriptions

noah_mendelsohn@us.ibm.com wrote:
> Xiaoshu Wang writes:
>
>   
>> Wouldn't it be more meaningful and precise to designate a GRDDL
>> mime-type and bound it transformation sheet to the same URI?
>>     
>
> Doesn't this make it impossible to have GRDDL for the GRDDL document 
> itself?  Granting that this may be a stretch in common usage, I think it's 
> a good test of architectural soundness that these edge cases work out.  I 
> think what we're seeing here:  mime types are closer to what RDDL calls a 
> "nature" than a "purpose".  You're using them for a "purpose", I think.
>   
So, GRDDL itself has the concept of "nature" or "purpose".  It follows 
that the GRDDL document itself can contain GRDDL reference, yes?  If so, 
why does it need to use MIME type anymore?

The context of Harry's use case is that an XML document's schema is 
fixed, which prohibited from including any internal <link> to external 
resource.  If there is no such restriction, there is no need for 
HTTP-LINK, obviously no need for MIME-type either.  I did not oppose 
internal link, which I said is useful. 

Obviously, we can imagine a scenario where GRDDL prohibits such link as 
well.  But, first, I have said before, we shouldn't design a system to 
cover corner cases because the consequence is that the system will 
eventually be too complicated to use.  Second, if the first version of 
GRDDL doesn't allow internal <link>, which means at the time there is no 
need to. If, later, there is such a need and GRDDL v1 cannot be extended 
(which make the usecase identical to Harry's usecase), we can always 
make a second version of GRDDL to do so.  Then, given a different MIME 
type, the problem can still be solved.  Either way, there is no need to 
have both Conneg and LINK, yes?

Regards,

Xiaoshu

Received on Monday, 31 March 2008 22:08:07 UTC