Re: binary/file resource metadata

My comments below.

Richard Cyganiak wrote:
> 
> 
> On 25 Jul 2008, at 03:19, carmen r wrote:
>> say you GET some /a.jpg
>>
>> and additionally you want to return RDF about this image (tags, 
>> geodata, comments, distillations from EXIF)
>>
>> i asked about a GETMETA and was told thats synonymous with HEAD.. only 
>> true if your RDF/metadata is all in the header.
>>
>> so there is a Link/N3 header proposed. that feels kinda nasty to me. 
>> id rather use RDF to express all the headers (gaining flexibility, and 
>> eliminating the need for this hack), than embed RDF inside a 
>> particular field (which has the bonus of backwards compatibiltiy with 
>> unmaintained UAs
>>
>>
>> what about Multipart-Mixed reponses. eg returning image/jpeg data, as 
>> well as a text/n3 file. can Tabulator or any tool grok these kind of 
>> responses ?
>>
>> does anyone have a preference for one technique over the other?
> 
> I think what people propose these days is resurrecting the Link header:
> 
>    Link: </a.jpg.rdf>;rel=meta
> 
> I'm not sure if this is the “hack” you refer to. To me, this looks nice 
> and clean and RESTful.

All being well, HTTP Link is indeed about to rise from the undead (for 
some of us it never really went away - Opera and Firefox both implement 
it as you can see at [1]).

Mark Nottingham has a draft at [2] that has been endorsed, at least for 
some uses, by the TAG (see the minutes of their discussion at [3]). I'm 
hoping that it will advance to RDF status in due course. The TAG 
discussion, loads of stuff on the HTTP and TAG mailing lists and a wiki 
page at [4] show that this subject has a _long_ history - and I very 
much doubt that any single solution will prevail - but HTTP Link looks 
like it does what you want it to do. Of course, POWDER will let you add 
triples to all your images at once so that should make it easier where 
you want to save time and effort (that's the idea anyway ;-))

All being well, POWDER will go to Last Call next week (I'm finalising 
the docs right now) and the one relevant to this discussion is at [5]. 
For now we've flagged using HTTP Link as a Feature at Risk (in case the 
RDF process trips up) but the signs are good.

HTH

Phil.



[1] http://www.fosi.org/archive/httplinktest/
[2] http://tools.ietf.org/html/draft-nottingham-http-link-header-02
[3] http://www.w3.org/2001/tag/2008/05/20-minutes#item06
[4] http://esw.w3.org/topic/FindingResourceDescriptions
[5] http://www.w3.org/TR/powder-dr/

-- 
Phil Archer
Chief Technical Officer,
Family Online Safety Institute
w. http://www.fosi.org/people/philarcher/

Received on Friday, 25 July 2008 10:06:04 UTC