W3C home > Mailing lists > Public > public-lod@w3.org > April 2010

Re: semantic pingback improvement request for foaf

From: Sebastian Tramp <tramp@informatik.uni-leipzig.de>
Date: Sat, 17 Apr 2010 11:18:05 +0200 (CEST)
To: Story Henry <henry.story@bblfish.net>
cc: public-lod community <public-lod@w3.org>, foaf-protocols@lists.foaf-project.org, Sören Auer <auer@informatik.uni-leipzig.de>, Philipp Frischmuth <pfrischmuth@googlemail.com>
Message-ID: <alpine.DEB.2.00.1004171035570.14286@fbywnevf4>
quote Story Henry (16.4.2010):

>> One of the main advantages of pingback (and thus Semantic Pingback) is 
>> the low entrance barrier for clients which want to trigger a 
>> pingback-request. This is done via a X-Pingback field in HTTP header of 
>> a pingback enabled resource. The auto discovery via header field is 
>> both simple and powerful. simple because a pingback clients does not 
>> need to use an XML parser (or an rdf parser)
> In the initial part described above this is true, and only if the HEAD 
> contains the X-Pingback link header.  (Otherwise you do need to parse 
> the content)

yep. but typical software like wordpress and other cms is using this 
header field since many years.

The ping:service property was intended more for foaf profiles, where the 
user is not able to manipulate the header himself (but its easy to put 
this line to your htacces ... Header set X-Pingback 
"http://pingback.aksw.org/" )

>> The suggested no.1 auto suggestion method for a resource is still the 
>> header field (this policy preserves backward compatibility, allowing 
>> wordpress users to ping a linked data resource).
>
> yes, the backward compatibility is also very useful, and should be kept.

>> According to your "Improved Semantic PingBack" approach (5). I 
>> understand your motivation to wipe-out the XMLRPC part of such a spec 
>> and maybe a POST request could be an alternative for such an RPC call
>
> I don't think that one should wipe out the XMLPRC piece in fact. I 
> should make that clearer in the proposed definition of the ping:service 
> relation. I was thinking one could extend it. Toby Inkster just made the 
> point very well in another thread on the foaf-protocols mailing list
>
>   http://markmail.org/message/xg4vfqitn2rqcrc7
>
> When you POST a content, you are in fact doing something like reverse 
> conneg. One error of the ping back service is to be entirely defined in 
> terms of a representation.

I think the idea of this definition was to have a extreme easy 
implementation complexity which is imho one of the key qualities of the 
pingback spec.

>> (however, this buries the compatibility to the blogosphere).
> It need not. We could have it so that ping back servers that are only 
> referred to via the service:pingback relation in rdf, can offer only the 
> new ping method.

This means that the property has a different interpretation in different 
contexts. as I said, ping:service could be used to query for a 
pingback server of a resource and equip the header field with the server 
resource.

> The other thing would be to have a new relation. In the case of 
> friending services in foaf, it is unlikely that people will have their 
> foaf served with X-Pingback headers.

If they want to accept pings from old pingback clients, they will. And its 
very easy to setup ...

>> But in order to be as lightweight as the original spec is, we should 
>> then introduce a new header-field as well to allow easy auto-discovery 
>> of such a Post-Pingback endpoint as well.
>>
>> Since the XMLRPC related header is named X-Pingback, I suggest to use 
>> X-Pingback-Post for this thus lifting post-pingbacks as an alternative 
>> request to xmlrpcs.
> XML-RPC sends the information to the server also via a POST, so I am not 
> sure that this would be the right way to distinguish them.

ok.

> I would be more in favour of calling the new relation Pingback without 
> the X, ...

> So I believe that Pingback needs to be fixed. The xml/rpc piece is just 
> wrong. The easiest way may be to start off with another relation
>
>  ping:pingback
>
> perhaps defined as I suggested, and then to find ways to tie this in 
> more carefully with the xml-rpc ping back service.

ok.

>> Semantic Pingback clients need to implement one of these methods while 
>> Semantic Pingback server should implement both to allow both request 
>> types and stay compatible.
>
> yes.

However, if a semantic pingback client implements only ping:pingback, it 
is not able to ping current blog-posts - and this is a very important 
use-case.

> Anyway, we are looking for a space to maintain the ontology. Is this 
> something you would be interested in adding to your semantic pingback 
> page?

yes of course.

S.Tramp

-- 
Sebastian Tramp - Department of Computer Science; University of Leipzig
Tel/Fax: +49 341 97 323-66/-29 http://bis.uni-leipzig.de/SebastianTramp
Received on Saturday, 17 April 2010 09:47:22 UTC

This archive was generated by hypermail 2.3.1 : Sunday, 31 March 2013 14:24:26 UTC