Minutes of this morning's call, August 2, 2009

["Denenberg, Ray" <rden@loc.gov>]

Present: Janifer, Ralph, Tony, Ray


1. Explain.
Janifer has provided a specification document (Word), 
     http://www.oasis-open.org/apps/org/workgroup/search-ws/download.php/38784/SRU%20explain%20document%202010_07_26.doc
and has also revised the XML Schema (.xsd), which is now at:
     http://www.oasis-open.org/apps/org/workgroup/search-ws/download.php/38786/SRU2-zeerex-2010-07-27.xsd

The new schema name for explain - in the OASIS context- is "SRU Explain. 
Discussion about the relationship of SRU Explain to Zeerex.  If you do a GET on an SRU 2.0 server, what do you get back, an SRU Explain or Zeerex record?  The consensus is that either should be acceptable. If a server has implemented 2.0 by ugrading from 1.x, but has not implemented any new features that would necessitate changing the Zeerex record, then it should not be required to change the Zeerex record to conform to an SRU Explain record. 

Change the protocol name, in SRU Explain, from "SRW/U" to "SRU"

SRU Explain will have a new (OASIS style) namespace. 

 responseFormats ("list of response formats supported") should probably be httpAccept and should list the values of httpAccept supported, and for each, a list of response formats that can be used in conjunction with that value. That list could be empty since a value of httpAccept could be supplied without a response format (and probably will be in most cases).  And for every pair (httpAccept value, responseFormat) a short name, that can be used to denote the pair.


2. Scan
A new draft of the Scan specification was issued,
     http://www.oasis-open.org/apps/org/workgroup/search-ws/download.php/38670/scan-july15-2010.doc
and a list of discussion points in the email:
    http://www.oasis-open.org/apps/org/workgroup/search-ws/email/archives/201007/msg00005.html


1.1 Review the data model.
No comments.

1.2 The processing model: is it sufficient or do we need the formality that we have for the SRU processing model.
No comments.

1.3 Diagnostic model.  Currently empty. 
Ralph will write it.

1.4 Explain model.  
Ralph not happy with Explain model, he will rewrite it.

1.5 Serialization model. Is scan strictly XML?
Spec will declare that it addresses XML only; other serializations are not precluded, but are out of scope. 
Also, explicitly out of scope is binding to lower layers, e.g. HTTP.

2.2 review description of parameters, particularly example in 2.2.2.
No discussion, not yet reviewed.

2.3 serialization. Should we use/adapt the SRU serialization model for XML/HTTP?
No.

3.3.1 whereInList. review.
Looks fine. 

For stylesheet: should we introduce rendering, client side vs. server, as we do for SRU?
No.

Extensions: should we write up an extensions section similar to SRU?
Add a section by adapting what we have for the current 1.2 spec.

Conformance: what should constitute conformance -  For a client? for a server?
Did not get around to discussing this.

Finally there was discussion about whether SRU and Scan are different protocols or part of the same protocols - that is, different operations, so there would be one protocol with three operations, SRU, scan, and Explain.  We didn't have time to explore this in depth but most likely the single protocol/multiple operation approach will be followed. 

We will tentatively hold our next call next week, August 9, if there is sufficient work during this week:  Ralph is tasked with the Scan document and will try to have a new draft by late Thursday.  If by friday there does not seem to be sufficient reason for a call August 9, it will be cancelled. In any case, we will hold a call in four weeks, August 30, and we will not hold a call either August 16 or August 23.