Re: ACTION-43 Draft use case for ordering

hello richard.

On 2013-03-26 01:59 , "Richard Cyganiak" <richard@cyganiak.de> wrote:
>On 25 Mar 2013, at 22:54, "Wilde, Erik" <Erik.Wilde@emc.com> wrote:
>> we cannot rely on client-side ordering,
>> because ordering is not necessarily based on anything *in* the
>>response. a
>> service could very well support ordering based on information that is
>>not
>> exposed in the response.
>That's true, but maybe not that critical, given that it's always easy for
>the server to add one more property to the container members to expose
>the value used for ordering.

good point. basically, it would be a generated sequence number that would
be there because RDF has no inherent concept of ordered data.

>>but the interesting question still remains: how
>> do we expose these capabilities, so that clients can actually send a
>> request that will result in the response as shown by richard?
>There's two issues: how do clients request a particular sort order. And
>how do servers expose the sort order in a response. The current LDP spec
>doesn't address the first question. That's ok with me. It's a version
>one. It ought to answer the second question if we support paging, for the
>reasons Steve stated in the initial mail of this thread.

i think it makes sense to first solve question two (how to represent
order). and yes, maybe we will not address question one in version one. in
AtomLand, question two was solved just because XML is ordered, and
question one never was solved on the Atom level, but various standards
evolved in specific communities such as OpenSearch.

cheers,

dret.

Received on Tuesday, 26 March 2013 20:00:57 UTC