[rus-wg] RUS Specification - operation results
John Ainsworth
john.ainsworth at manchester.ac.uk
Sat Apr 9 01:44:48 CDT 2005
I think the use case Sven presents is valid, but I don't like the idea
of having so many parallel operations in the interface though. I think
the operation Steven describes is all that would be needed.
John
On 7 Apr 2005, at 17:09, Sven van den Berghe wrote:
> Steven,
>
> I can see some cases where the purpose of a query is only to get one
> (or
> more) RUSIds for further processing (e.g. updates) so returning a full
> record is excessive.
>
> Sven
>
>
> "Steven Newhouse" <sjn5 at doc.ic.ac.uk> wrote:
>
>>> The extract/fetch operations seem to return one or more full
>>> RUSUsageRecords, would it be sensible to add parallel operations
>>> that return
>>> just the ids?
>>
>> I think you identify a need for just a single additional operation.
>>
>> RUS::extractRecordsRUSId
>> INPUT: XPath/XQuery
>> OUTPUT: List of RUSIds
>>
>> The existing operation RUS::extractSpecificRecords could then be used
>> to
>> extract the relevant records.
>>
>> Following the initial extraction of the RUS Ids what would the client
>> do
>> with them? As there is no meta-data other than the ID one of the few
>> things would seem to iterate over them to retrieve specific records
>> in turn.
>>
>> The list of RUSIds from one search operation could be compared to that
>> of another... but then the two queries could be merged into one to
>> yield
>> the required (from the combined query) records in one go.
>>
>>> The would be more efficient when the purpose is just to
>>> determine existence and not necessarily examine contents.
>>
>> Do you have additional use cases in mind as to what you would do with
>> the resulting RUSIds?
>>
>>> On a related note, I don't know XPath/XQuery well but I assume that
>>> it is
>>> possible to do something like "extract all records that have been
>>> created in
>>> the last month". Is this correct?
>>
>> As the initial timestamp element in the RecordHistory follows a
>> standard
>> format it should be searchable by XPath/Xquery... but I'm not an
>> expert
>> in this.
>>
>> Can anyone clarify what XPath/XQuery can dow with timestamps?
>>
>> Steven
>
> --
> Sven van den Berghe
> Fujitsu Laboratories of Europe
> +44 208 606 4651
>
>
>
>
More information about the rus-wg
mailing list