Re: primer review part 2

Jeremy Carroll wrote:
>> I don't see that, can you explain? The HTML document at
>> http://research.talis.com/2005/erdf/profile contains the following
>> markup:
>>
>>         <p>This following link provides the statement : <a
>>         rel="profileTransformation"
>>  
>> href="http://purl.org/NET/erdf/extract-rdf.xsl">extract-rdf.xsl</a>, <a
>>         rel="profile"
>> href="http://purl.org/NET/erdf/profile">profile</a></p>
> 
> 
> Aah.
> I might have made a mistake in the rationalization of the library 
> glean-profile function.
> 
> The idiom rel="profile" is intended to indicate the profile to which the 
> profileTransformation applies?
> 


Looking in the history of glean-profile, at one point it supported an 
idea of subprofile, which includes links annotated with rel="profile". I 
removed this because it went further than seemed appropriate in a 
conservative library function (e.g. subprofiles are not defined anywhere 
...) The actual code would not have applied to the ERDF profile: it only 
ever worked because implementations did not follow 5.1.3 of RFC 3986, 
and use the URL before the redirect rather than after.


However, the use-case for glean-profile of a profile that can be 
retrieved by a redirected URL seems normal. Ideally such a profile 
should generate two profile transformation triples, e.g. in this case:

<http://purl.org/NET/erdf/profile>
grddl:profileTransformation
<http://purl.org/NET/erdf/profile> .

<http://research.talis.com/2005/erdf/profile>
grddl:profileTransformation
<http://purl.org/NET/erdf/profile> .

since either URL might be used as a metadata profile name in an HTML 
document, and it would be useful for both to work.

I'll try modifying the library code and see what happens (not today though).

Jeremy



-- 
Hewlett-Packard Limited
registered Office: Cain Road, Bracknell, Berks RG12 1HN
Registered No: 690597 England

Received on Thursday, 7 June 2007 16:43:51 UTC