W3C home > Mailing lists > Public > public-lod@w3.org > February 2013

Re: Content negotiation for Turtle files

From: Ivan Herman <ivan@w3.org>
Date: Wed, 6 Feb 2013 12:16:12 +0100
Cc: "Bernard Vatant" <bernard.vatant@mondeca.com>, "Linking Open Data" <public-lod@w3.org>
Message-Id: <53C6EEA9-6618-4040-ADAA-1CB5AA81E71A@w3.org>
To: chris@codex.net.au

On Feb 6, 2013, at 10:56 , "Chris Beer" <chris@codex.net.au> wrote:

> Bernard, Ivan
> 
> (At last! Something I can speak semi-authoritatively on ;P )
> 
> @ Bernard - no - there is no reason to go back if you do not want to, and
> every reason to serve both formats plus more.
> 
> Your comment about UA's complaining about a content negotiation issue is
> key to what you're trying to do here. I'd like to provide some clear
> guidance or suggestions back, but first, if possible, can you please post
> the http request headers for the four (and any others you have) user
> agents you've used to attempt to request your rdf+xml files and which have
> either choked or accepted the .ttl file. Extra points if you can also post
> the server's response headers.
> 
> @ Ivan - while I wince a little at the trick - the question comes down to
> the same thing - what is the http response header that is sent back to the
> client -

See my separate answer to Bernard.


> would be interested to see if in fact what you're doing ISN'T a
> "trick" but in fact a compliant way to approach this.
> 

Well, o.k. The term 'trick' may not be well chosen; it is probably the standard way of doing this on Apache.

> Personally I think you shouldn't actually need to resort to using .var
> (which is Apache specific) when what is essentially a content negotiation
> issue can simply be configured properly at the server level and thus a
> single approach could be used by IIS, Apache, nginx etc.
> 

That, unfortunately, I do not know. 

Ivan


> Look forward to the responses (excuse the pun)
> 
> Cheers
> 
> Chris
> 
> --------------
> 
> Chris Beer
> Manager - Online Services
> Department of Regional Australia, Local Government, Arts and Sport
> 
>> Bernard,
>> 
>> (forget my W3C hat, I am not authoritative on Apache tricks, for
>> example...)
>> 
>> When I put up a vocabulary onto www.w3.org/ns/, for example, I publish it
>> both in ttl and rdf/xml. Actually, we also publish the file in HTML+RDFa
>> (which very often is the master copy and I convert it into ttl and rdf/xml
>> before publishing). Additionally, we put there a .var file. This is the
>> .var file for the http://www.w3.org/ns/r2rml:
>> 
>> r2rml.var
>> ---------
>> URI: r2rml
>> 
>> URI: r2rml.html
>> Content-Type: text/html
>> 
>> URI: r2rml.rdf
>> Content-Type: application/rdf+xml; qs=0.4
>> 
>> URI: r2rml.ttl
>> Content-Type: text/turtle; qs=0.5
>> 
>> that seems to work well, at least I have not heard complaints:-)
>> 
>> One can do a further trick by adding to .htaccess entries to convert, say,
>> r2rml.html to r2rml.ttl on the fly; I did not do that to reduce the load
>> on our servers.
>> 
>> There is somewhere a flag in the apache configuration allowing apache to
>> handle these .var files; I am not sure it is there by default.
>> 
>> I hope this helps
>> 
>> Ivan
>> 
>> 
>> 
>> 
>> 
>> On Feb 6, 2013, at 24:49 , Bernard Vatant <bernard.vatant@mondeca.com>
>> wrote:
>> 
>>> Hello all
>>> 
>>> Back in 2006, I thought had understood with the help of folks around
>>> here, how to configure my server for content negotiation at lingvoj.org.
>>> Both vocabulary and instances were published in RDF/XML.
>>> 
>>> I updated the ontology last week, and since after years of happy living
>>> with RDF/XML people eventually convinced that it was a bad, prehistoric
>>> and ugly syntax, I decided to be trendy and published the new version in
>>> Turtle at http://www.lingvoj.org/ontology_v2.0.ttl
>>> 
>>> The vocabulary URI is still the same : http://www.lingvoj.org/ontology,
>>> and the namespace  http://www.lingvoj.org/ontology# (cool URI don't
>>> change)
>>> 
>>> Then I turned to Vapour to test this new publication, and found out that
>>> to be happy with the vocabulary URI it has to find some answer when
>>> requesting application/rdf+xml. But since I have no more RDF/XML file
>>> for this version, what should I do?
>>> I turned to best practices document at
>>> http://www.w3.org/TR/swbp-vocab-pub, but it does not provide examples
>>> with Turtle, only RDF/XML.
>>> 
>>> So I blindly put the following in the .htaccess : AddType
>>> application/rdf+xml .ttl
>>> I found it a completely stupid and dirty trick ... but amazigly it makes
>>> Vapour happy.
>>> 
>>> But now Firefox chokes on http://www.lingvoj.org/ontology_v2.0.ttl
>>> because it seems to expect a XML file. Chrome has not this issue.
>>> The LOV-Bot says there is a content negotiation issue and can't get the
>>> file. So does Parrot.
>>> 
>>> I feel dumb, but I'm certainly not the only one, I've stumbled upon a
>>> certain number of vocabularies published in Turtle for which the conneg
>>> does not seem to be perfectly clear either.
>>> 
>>> What do I miss, folks? Should I forget about it, and switch back to good
>>> ol' RDF/XML?
>>> 
>>> Bernard
>>> 
>>> --
>>> Bernard Vatant
>>> Vocabularies & Data Engineering
>>> Tel :  + 33 (0)9 71 48 84 59
>>> Skype : bernard.vatant
>>> Blog : the wheel and the hub
>>> --------------------------------------------------------
>>> Mondeca
>>> 3 cité Nollez 75018 Paris, France
>>> www.mondeca.com
>>> Follow us on Twitter : @mondecanews
>>> ----------------------------------------------------------
>>> 
>>> Meet us at Documation in Paris, March 20-21
>>> 
>>> 
>> 
>> 
>> ----
>> Ivan Herman, W3C Semantic Web Activity Lead
>> Home: http://www.w3.org/People/Ivan/
>> mobile: +31-641044153
>> FOAF: http://www.ivan-herman.net/foaf.rdf
>> 
>> 
>> 
>> 
>> 
>> 
> 
> 


----
Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
mobile: +31-641044153
FOAF: http://www.ivan-herman.net/foaf.rdf







Received on Wednesday, 6 February 2013 11:16:40 UTC

This archive was generated by hypermail 2.3.1 : Sunday, 31 March 2013 14:24:46 UTC