W3C home > Mailing lists > Public > public-rdf-dawg@w3.org > October to December 2009

Re: REST and HTTP Update

From: Steve Harris <steve.harris@garlik.com>
Date: Wed, 7 Oct 2009 20:26:11 +0100
Message-Id: <34B9EB84-028D-4383-B05D-0BB4AB2709EB@garlik.com>
To: "public-rdf-dawg@w3.org Group" <public-rdf-dawg@w3.org>
On 7 Oct 2009, at 19:09, Kjetil Kjernsmo wrote:


> However, my main point is that we shouldn't have this discussion at  
> all
> right now, this should have been done on a time-permitting basis only.
> There are so many good things that we have ruled out (I still mourn  
> the
> loss of the fulltext index ;-) ), and this is such a marginal  
> improvement
> over "pure REST + the Update Language", it shouldn't be a topic at  
> all at
> this point.

That's not been our experience. Our store supports both the obvious  
REST interface and the $endpoint + urlencode($graph) thing. I think we  
probably use them both about equally, and all our updates are done  
with PUT or POST.

However, among the other people using our store, I don't know of  
anyone using the obvious REST approach. For whatever reason* people  
seem to prefer the urlencode form. The half dozen or so 3rd party  
libraries all seem to use the urlencode form.

The ?graph= thing would also work OK for us BTW.

* I guess it might be related to use the use reverse proxies, which  
seem very popular. You can't really use the obvious REST interface  
with a proxy.

> That is not to say that it isn't useful, it is (our current solution  
> at
> work can't even do pure REST), and I can see that it is unfriendly  
> to parse
> RDF/XML and serialize to Turtle to use the Language. But still, I  
> think it
> should be on a time-permitting basis only, because we can spend huge
> amounts of time and delay the specs by arguing over this, even without
> being very much in disagreement (I don't think we are).

It's not just parsing RDF/XML into Turtle, you have to parse it into  
SPARQL/Update syntax, extra {}s, PREFIX not @prefix, different qname  
rules(?), so it's really another syntax form.

> So, how about a tiny little paragraph about this, and leave it at  
> that,
> with a note to the effect that it is an "at risk" feature?
>> Consider the 'opacity axiom' [1] which (I believe) is often  
>> associated
>> with claims that URIs which require an intermediate agent to parse  
>> their
>> lexical form violates a best practice:
> Yup, but there is no intermediate agent in Steve's proposal, is there?

I don't think so.

- Steve

Steve Harris
Garlik Limited, 2 Sheen Road, Richmond, TW9 1AE, UK
+44(0)20 8973 2465  http://www.garlik.com/
Registered in England and Wales 535 7233 VAT # 849 0517 11
Registered office: Thames House, Portsmouth Road, Esher, Surrey, KT10  
Received on Wednesday, 7 October 2009 19:26:46 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:00:57 UTC