Date:         Thu, 1 Nov 2001 09:03:37 -0000
Reply-To:     "Z39.50 Next-Generation Initiative" <[log in to unmask]>
Sender:       "Z39.50 Next-Generation Initiative" <[log in to unmask]>
From:         Matthew Dovey <[log in to unmask]>
Subject:      Re: SOAP packet content questions
Comments: To: "Z39.50 Next-Generation Initiative" <[log in to unmask]>
Content-Type: text/plain; charset="us-ascii"

Personally I thought we had simplified things quite a lot already but...


> Question 1: What is SRW? Should I be saying that instead of ZNG?

Search/Retrieve for the Web. Ray wants ZNG or rather ZiNG to refer to
the umbrella of project looking at new ways of doing Z39.50 like things
(so SOAP, ZOOM etc.)
 
> Question 2: Why are Bib-1 diagnostics in the WSDL file? Why not
> make it a bit more human friendly and just include a string
description.
> Reduces learning curve etc. Lets face - most of the time if something
> goes wrong you just want to say why.

Just because we are now using XML it doesn't change the fact that in
many cases it wo'n't be a human reading the XML directly but a machine
process which may need to take some action on discovering a certain
error. String only values for diagnostics could easily lead to different
implementations returning different string messages for the same error
(or class of error). In any case the diagnostic structure in the WSDL
includes a diagnostic code (number) and a diagnostic message (string) so
meets both cases (human processing the XML and machine processing the
XML).

> Question 3: Is there a list of responseSchema values people are
> actually using? What is the current thinking here?

Not yet. There was some discussion that we needed to pass back the
original query etc. in the returned XML. This has also resurfaced a few
times on this list. I disagree on the KISS principle and because most
clients will know what they sent anyway. However, there is meant to be a
schema as an extension to the current one which does include this. The
idea here is that a XML browser could just do an XSLT tranform to
display the results. I don't think that this schema exists yet.

> Question 4: Is record schema the same as record syntax? Or different?

Not quite. In traditional Z39.50 we appear to have mixed up record
syntax and record schema. I would argue that ISO 2709 (Binary Marc) and
XML are record syntaxes, whilst MARC21, UKMARC are schemas/flavours of
ISO2709, and METS, ONIX etc. are schema/flavours of XML. In SRW, we only
use one record syntax namely XML, but need to support different XML
schemas.

However, in practice this does the same thing as the record syntax
specification in Z39.50.

> Question 5: In order to simplify things, is it worth dropping all
> diagnostics codes, additional information, etc and using SOAP
> fault codes instead (ie standard SOAP mechanism). You only get one
> error per packet, but is this a problem in practice?

Possibly. This does raise the question on how surrogate diagnostics are
handled. In the current form I used the same structure for both
non-surrogate and surrogate diagnostics (again KISS principle). 

Matthew