Re: [search-ws] RE: APP?

[Farrukh Najmi <farrukh@wellfleetsoftware.com>]

I assume by APP you mean Atom Publishing Protocol 
<http://en.wikipedia.org/wiki/Atom_Publishing_Protocol>?

If so, the suggestion is probably to use the ATOM feed format for search 
results instead of inventing our own.
I strongly support this suggestion and have logged it under as P1 issue 
farrukh-6 at:

<http://wiki.oasis-open.org/search-ws/issues>

The discussion thread on that proposal is at:

<http://www.oasis-open.org/apps/org/workgroup/search-ws/email/archives/200710/msg00068.html>

Forgive me if I misunderstood the issue. Thanks.

LeVan,Ralph wrote:
>> From: Identifiers Working Group List
>> [mailto:IDENTIFIERS-L@OCLC.ORG]
>> On Behalf Of Houghton,Andrew
>> Sent: Friday, November 09, 2007 10:50 AM
>> To: identifiers-l
>> Subject: RE: APP?
>>
>> Since the OASIS committee is already redefining SRU and is
>> not tied down to maintain backward compatibility, then
>> using an APP response would be worth while.  However,
>> not being on the committee, its not clear why they
>> would start with SRU as a basis and not APP.
>>     
>
> Because there are no signs of any searching capability in APP?
>
>   
>> One of the downsides to SRU is there is no equivalent APP
>> service document.  So I just cannot query a URL and find
>> out about all the collections that an SRU server supports.
>>     
>
> True.  That was in intentional decision when we created SRU.  We were
> reacting to the overly complex Explain mechanism of Z39.50 that needed
> to support the ability to describe how a client might search across
> multiple databases.  That turned into a nightmare.
>
> In SRU we decided to just describe the collections themselves.  I agree
> that a simple list of the collections would be useful.  In fact, my SRU
> server provides just such a list because I found it useful.  You can be
> sure that it is being discussed in the OASIS group.
>
>   
>> The Explain request is for a specific collection and its,
>> IMHO, worthless since while it tells you the names of the
>> indexes, it doesn't tell you whether the index is a phrase
>> or word index and which operators are supported on that
>> index.  So if you are trying to write a generic SRU client
>> to talk to unknown SRU servers in an SOA context, it
>> just doesn't work.
>>     
>
> I suspect you are reacting to my own implementation of SRU.  I'm sure
> the Explain experts on the SRU list could tell us how to specify what
> operations are supported on a particular index.
>   


-- 
Regards,
Farrukh Najmi

Web: http://www.wellfleetsoftware.com